Home > Blogs > VMware VROOM! Blog > Tag Archives: vsphere

Tag Archives: vsphere

Virtualized Storage Performance: RAID Groups versus Storage pools

RAID, a redundant array of independent disks, has traditionally been the foundation of enterprise storage. Grouping multiple disks into one logical unit can vastly increase the availability and performance of storage by protecting against disk failure, allowing greater I/O parallelism, and pooling capacity. Storage pools similarly increase the capacity and performance of storage, but are easier to configure and manage than RAID groups.

RAID groups have traditionally been regarded as offering better and more predictable performance than storage pools. Although both technologies were developed for magnetic hard disk drives (HDDs), solid-state drives (SSDs), which use flash memory, have become prevalent. Virtualized environments are also common and tend to create highly randomized I/O given the fact that multiple workloads are run simultaneously.

We set out to see how the performance of RAID group and storage pool provisioning methods compare in today’s virtualized environments.

First, let’s take a closer look at each storage provisioning type.

RAID Groups

A RAID group unifies a number of disks into one logical unit and distributes data across multiple drives. RAID groups can be configured with a particular protection level depending on the performance, capacity, and redundancy needs of the environment. LUNs are then allocated from the RAID group. RAID groups typically contain only identical drives, and the maximum number of disks in a RAID group varies by system model but is generally below fifty. Because drives typically have well defined performance characteristics, the overall RAID group performance can be calculated as the performance of all drives in the group minus the RAID overhead. To provide consistent performance, workloads with different I/O profiles (e.g., sequential vs. random I/O) or different performance needs should be physically isolated in different RAID groups so they do not share disks.

Storage Pools

Storage pools, or simply ‘pools’, are very similar to RAID groups in some ways. Implementation varies by vendor, but generally pools are made up of one or more private RAID groups, which are not visible to the user, or they are composed of user-configured RAID groups which are added manually to the pool. LUNs are then allocated from the pool. Storage pools can contain up to hundreds of drives, often all the drives in an array. As business needs grow, storage pools can be easily scaled up by adding drives or RAID groups and expanding LUN capacity. Storage pools can contain multiple types and sizes of drives and can spread workloads over more drives for a greater degree of parallelism.

Storage pools are usually required for array features like automated storage tiering, where faster SSDs can serve as a data cache among a larger group of HDDs, as well as other array-level data services like compression, deduplication, and thin provisioning. Because of their larger maximum size, storage pools, unlike RAID groups, can take advantage of vSphere 6 maximum LUN sizes of 64TB.

We used two benchmarks to compare the performance of RAID groups and storage pools: VMmark, which is a virtualization platform benchmark, and I/O Analyzer with Iometer, which is a storage microbenchmark.  VMmark is a multi-host virtualization benchmark that uses diverse application workloads as well as common platform level workloads to model the demands of the datacenter. VMs running a complete set of the application workloads are grouped into units of load called tiles. For more details, see the VMmark 2.5 overview. Iometer places high levels of load on the disk, but does not stress any other system resources. Together, these benchmarks give us both a ‘real-world’ and a more focused perspective on storage performance.

VMmark Testing

Array Configuration

Testing was conducted on an EMC VNX5800 block storage SAN with Fibre Channel. This was one of the many storage solutions which offered both RAID group and storage pool technologies. Disks were 200GB single-level cell (SLC) SSDs. Storage configuration followed array best practices, including balancing LUNs across Storage Processors and ensuring that RAID groups and LUNs did not span the array bus. One way to optimize SSD performance is to leave up to 50% of the SSD capacity unutilized, also known as overprovisioning. To follow this best practice, 50% of the RAID group or storage pool was not allocated to any LUN. Since overprovisioning SSDs can be an expensive proposition, we also tested the same configuration with 100% of the storage pool or RAID group allocated.

RAID Group Configuration

Four RAID 5 groups were used, each composed of 15 SSDs. RAID 5 was selected for its suitability for general purpose workloads. RAID 5 provides tolerance against a single disk failure. For best performance and capacity, RAID 5 groups should be sized to multiples of five or nine drives, so this group maintains a multiple of the preferred five-drive count. One LUN was created in each of the four RAID groups. The LUN was sized to either 50% of the RAID group (Best Practices) or 100% (Fully Allocated). For testing, the capacity of each LUN was fully utilized by VMmark virtual machines and randomized data.

            RAID Group Configuration VMmark Storage Comparison        VMmark Storage Pool Configuration Storage Comparison

Storage Pool Configuration

A single RAID 5 Storage Pool containing all 60 SSDs was used. Four thick LUNs were allocated from the pool, meaning that all of the storage space was reserved on the volume. LUNs were equivalent in size and consumed a total of either 50% (Best Practices) or 100% (Fully Allocated) of the pool capacity.

Storage Layout

Most of the VMmark storage load was created by two types of virtual machines: database (DVD Store) and mail server (Microsoft Exchange). These virtual machines were isolated on two different LUNs. The remaining virtual machines were spread across the remaining two LUNs. That is, in the RAID group case, storage-heavy workloads were physically isolated in different RAID groups, but in the storage pool case, all workloads shared the same pool.

Systems Under Test: Two Dell PowerEdge R720 servers
Configuration Per Server:  
     Virtualization Platform: VMware vSphere 6.0. VMs used virtual hardware version 11 and current VMware Tools.
     CPUs: Two 12-core Intel® Xeon® E5-2697 v2 @ 2.7 GHz, Turbo Boost Enabled, up to 3.5 GHz, Hyper-Threading enabled.
     Memory: 256GB ECC DDR3 @ 1866MHz
     Host Bus Adapter: QLogic ISP2532 DualPort 8Gb Fibre Channel to PCI Express
     Network Controller: One Intel 82599EB dual-port 10 Gigabit PCIe Adapter, one Intel I350 Dual-Port Gigabit PCIe Adapter

Each configuration was tested at three different load points: 1 tile (the lowest load level), 7 tiles (an approximate mid-point), and 13 tiles, which was the maximum number of tiles that still met Quality of Service (QoS) requirements. All datapoints represent the mean of two tests of each configuration.

VMmark Results

RAID Group vs. Storage Pool Performance comparison using VMmark benchmark

Across all load levels tested, the VMmark performance score, which is a function of application throughput, was similar regardless of storage provisioning type. Neither the storage type used nor the capacity allocated affected throughput.

VMmark 2.5 performance scores are based on application and infrastructure workload throughput, while application latency reflects Quality of Service. For the Mail Server, Olio, and DVD Store 2 workloads, latency is defined as the application’s response time. We wanted to see how storage configuration affected application latency as opposed to the VMmark score. All latencies are normalized to the lowest 1-tile results.

Storage configuration did not affect VMmark application latencies.

Application Latency in VMmark Storage Comparison RAID Group vs Storage Pool

Lastly, we measured read and write I/O latencies: esxtop Average Guest MilliSec/Write and Average Guest MilliSec/Read. This is the round trip I/O latency as seen by the Guest operating system.

VMmark Storage Latency Storage Comparison RAID Group vs Storage Pool

No differences emerged in I/O latencies.

I/O Analyzer with Iometer Testing

In the second set of experiments, we wanted to see if we would find similar results while testing storage using a synthetic microbenchmark. I/O Analyzer is a tool which uses Iometer to drive load on a Linux-based virtual machine then collates the performance results. The benefit of using a microbenchmark like Iometer is that it places heavy load on just the storage subsystem, ensuring that no other subsystem is the bottleneck.


Testing used a VNX5800 array and RAID 5 level as in the prior configuration, but all storage configurations spanned 9 SSDs, also a preferred drive count. In contrast to the prior test, the storage pool or RAID group spanned an identical number of disks, so that the number of disks per LUN was the same in both configurations. Testing used nine disks per LUN to achieve greater load on each disk.

The LUN was sized to either 50% or 100% of the storage group. The LUN capacity was fully occupied with the I/O Analyzer worker VM and randomized data.  The I/O Analyzer Controller VM, which initiates the benchmark, was located on a separate array and host.

Storage Configuration Iometer with Storage Pool and RAID Group

Testing used one I/O Analyzer worker VM. One Iometer worker thread drove storage load. The size of the VM’s virtual disk determines the size of the active dataset, so a 100GB thick-provisioned virtual disk on VMFS-5 was chosen to maximize I/O to the disk and minimize caching. We tested at a medium load level using a plausible datacenter I/O profile, understanding, however, that any static I/O profile will be a broad generalization of real-life workloads.

Iometer Configuration

  • 1 vCPU, 2GB memory
  • 70% read, 30% write
  • 100% random I/O to model the “I/O blender effect” in a virtualized environment
  • 4KB block size
  • I/O aligned to sector boundaries
  • 64 outstanding I/O
  • 60 minute warm up period, 60 minute measurement period
Systems Under Test: One Dell PowerEdge R720 server
Configuration Per Server:  
     Virtualization Platform: VMware vSphere 6.0. Worker VM used the I/O Analyzer default virtual hardware version 7.
     CPUs: Two 12-core Intel® Xeon® E5-2697 v2 @ 2.7 GHz, Turbo Boost Enabled, up to 3.5 GHz, Hyper-Threading enabled.
     Memory: 256GB ECC DDR3 @ 1866MHz
     Host Bus Adapter: QLogic ISP2532 DualPort 8Gb Fibre Channel to PCI Express

Iometer results

Iometer Latency Results Storage Comparison RAID Group vs Storage PoolIometer Throughput Results Storage Comparison RAID Group vs Storage Pool

In Iometer testing, the storage pool showed slightly improved performance compared to the RAID group, and the amount of capacity allocated also did not affect performance.

In both our multi-workload and synthetic microbenchmark scenarios, we did not observe any performance penalty of choosing storage pools over RAID groups on an all-SSD array, even when disparate workloads shared the same storage pool. We also did not find any performance benefit at the application or I/O level from leaving unallocated capacity, or overprovisioning, SSD RAID groups or storage pools. Given the ease of management and feature-based benefits of storage pools, including automated storage tiering, compression, deduplication, and thin provisioning, storage pools are an excellent choice in today’s datacenters.

Scaling Web 2.0 Applications using Docker containers on vSphere 6.0

by Qasim Ali

In a previous VROOM post, we showed that running Redis inside Docker containers on vSphere adds little to no overhead and observed sizeable performance improvements when scaling out the application when compared to running containers directly on the native hardware. This post analyzes scaling Web 2.0 applications using Docker containers on vSphere and compares the performance of running Docker containers on native and vSphere. This study shows that Docker containers add negligible overhead when run on vSphere, and also that the performance using virtual machines is very close to native and, in certain cases, slightly better due to better vSphere scheduling and isolation.

Web 2.0 applications are an integral part of Enterprise and small business IT offerings. We use the CloudStone benchmark, which simulates a typical Web 2.0 technology use in the workplace for our study [1] [2]. It includes a Web 2.0 social-events application (Olio) and a client implemented using the Faban workload generator [3]. It is an open source benchmark that simulates activities related to social events. The benchmark consists of three main components: a Web server, a database backend, and a client to emulate real world accesses to the Web server. The overall architecture of CloudStone is depicted in Figure 1.

Figure 1: CloudStone architecture

The benchmark reports latency for various user actions. These metrics were compared against a fixed threshold. Studies indicate that users are less likely to visit a Web site if the response time is greater than 250 milliseconds [4]. This number can be used as an upper bound for latency for frequent operations (Home-Page, TagSearch, EventDetail, and Login). For the less frequent operations (AddEvent, AddPerson, and PersonDetail), a less restrictive threshold of 500 milliseconds can be used. Table 1 shows the exact mix/frequency of various operations.

Operation Number of Operations Mix
HomePage 141908 26.14%
Login 55473 10.22%
TagSearch 181126 33.37%
EventDetail 134144 24.71%
PersonDetail 14393 2.65%
AddPerson 4662 0.86%
AddEvent 11087 2.04%

Table 1: CloudStone operations frequency for 1500 users

Benchmark Components and Experimental Set up

The test system was installed with a CloudStone implementation of a MySQL database, NGINX Web server with PHP scripts, and a Tomcat application server provided by the Faban harness. The default configuration was used for the workload generator. All components of the application ran on a single host, and the client ran in a separate virtual machine on a separate host. Both hosts were connected using a direct link between a pair of 10Gbps NICs. One client-server pair provided a single, independent CloudStone instance. Scaling was achieved by running additional instances of CloudStone.

Deployment Scenarios

We used the following three deployment scenarios for this study:

  • Native-Docker: One or more CloudStone instances were run inside Docker containers (2 containers per CloudStone instance: one for the Web server and another for the database backend) running on the native OS.
  • VM: CloudStone instances were run inside one or more virtual machines running on vSphere 6.0; the guest OS is the same as the native scenario.
  • VM-Docker: CloudStone instances were run inside Docker containers that were running inside one or more virtual machines.

Hardware/Software/Workload Configuration

The following are the details about the hardware and software used in the various experiments discussed in the next section:

Server Host:

  • Dell PowerEdge R820
  • CPU: 4 x Intel® Xeon® CPU E5-4650 @ 2.30GHz (32 cores, 64 hyper-threads)
  • Memory: 512GB
  • Hardware configuration: Hyper-Threading (HT) ON, Turbo-boost ON, Power policy: Static High (that is, no power management)
  • Network: 10Gbps
  • Storage: 7 x 250GB 15K RPM 4Gb SAS  Disks

Client Host:

  • Dell PowerEdge R710
  • CPU: 2 x Intel® Xeon® CPU X5680 @ 3.33GHz (12 cores, 24 hyper-threads)
  • Memory: 144GB
  • Hardware configuration: HT ON, Turbo-boost ON, Power policy: Static High (that is, no power management)
  • Network: 10Gbps
  • Client VM: 2-vCPU 4GB vRAM

Host OS:

  • Ubuntu 14.04.1
  • Kernel 3.13

Docker Configuration:

  • Docker 1.2
  • Ubuntu 14.04.1 base image
  • Host volumes for database and images
  • Configured with host networking to avoid Docker NAT overhead
  • Device mapper as the storage backend driver


  • VMware vSphere 6.0 (pre-release build)

VM Configurations:

  • Single VM: An 8-vCPU 4GB VM ( Web server and database running in a single VM)
  • Two VMs: One 6-vCPU 2GB Web server VM and one 2-vCPU 2GB database VM (CloudStone instance running in two VMs)
  • Scale-out: Eight 8-vCPU 4GB VMs

Workload Configurations:

  • The NGINX Web server was configured with 4 worker processes and 4096 connections per worker.
  • PHP was configured with a maximum number of 16 child processes.
  • The Web server and the database were preconfigured with 1500 users per CloudStone instance.
  • A runtime of 30 minutes with a 5 minute ramp-up and ramp-down periods (less than 1% run-to-run  variation) was used.


First, we ran a single instance of CloudStone in the various configurations mentioned above. This was meant to determine the raw overhead of Docker containers running on vSphere vs. the native configuration, eliminating scheduling differences. Second, we picked the configuration that performed best in a single instance and scaled it out to run multiple instances.

Figure 2 shows the mean latency of the most frequent operations and Figure 3 shows the mean latency of less frequent operations.

Figure 2: Results of single instance CloudStone experiments for frequently used operations


Figure 3: Results of single instance CloudStone experiments for less frequently used

We configured the benchmark to use a single VM and deployed the Web server and database applications in it (configuration labelled VM-1VM in Figure 2 and 3). We then ran the same workload in Docker containers in a single VM (VMDocker-1VM). The latencies are slightly higher than native, which is expected due to some virtualization overhead. However, running Docker containers on a VM showed no additional overhead. In fact, it seems to be slightly better. We believe this might be due to the device mapper using twice as much page cache as the VM (device mapper uses a loopback device to mount the file system and, hence, data ends up being cached twice in the buffer cache). We also tried AuFS as a storage backend for our container images, but that seemed to add some CPU and latency overhead, and, for this reason, we switched to device mapper. We then configured the VM to use the vSphere Latency Sensitivity feature [5] (VM-1VM-lat and VMDocker-1VM-lat labels). As expected, this configuration reduced the latencies even further because each vCPU got exclusive access to a core and this reduced scheduling overhead.  However, this feature cannot be used when the VM (or VMs) has more vCPUs than the number of cores available on the system (that is, the physical CPUs are over-committed) because each vCPU needs exclusive access to a core.

Next, we configured the workload to use two VMs, one for the Web server and the other for the database application. This configuration ended up giving slightly higher latencies because the network packets have to traverse the virtualization layer from one VM to the other, while in the prior experiments they were confined within the same VM.

Finally, we scaled out the CloudStone workload with 12,000 users by using eight 8-vCPU VMs with 1500 users per instance. The VM configurations were the same as the VM-1VM and VMDocker-1VM cases above. The average system CPU core utilization was around 70-75%, which is the typical average CPU utilization for latency sensitive workloads because it allows for headroom to absorb traffic bursts. Figure 4 reports mean latencies of all operations (latencies were averaged across all eight instances of CloudStone for each operation), while Figure 5 reports the 90th percentile latencies (the benchmark reports these latencies in 20 millisecond granularity as evident from Figure 5.)


Figure 4: Scale-out experiments using eight instances of CloudStone (mean latency)


Figure 5: Scale-out experiments using eight instances of CloudStone (90th percentile latency)

The latencies shown in Figure 4 and 5 are well below the 250 millisecond threshold. We observed that the latencies on vSphere are very close to native or, in certain cases, slightly better than native (for example, Login, AddPerson and AddEvent operations). The latencies were better than native due to better vSphere scheduling and isolation, resulting in better cache/memory locality. We verified this by pinning container instances on specific sockets and making the native scheduler behavior similar to vSphere. After doing that, we observed that latencies in the native case got better and they were similar or slightly better than vSphere.

Note: Introducing artificial affinity between processes and cores is not a recommended practice because it is error-prone and can, in general, lead to unexpected or suboptimal results.


VMs and Docker containers are truly “better together.” The CloudStone scale-out system, using out-of-the-box VM and VM-Docker configurations, clearly achieves very close to, or slightly better than, native performance.


[1] W. Sobel, S. Subramanyam, A. Sucharitakul, J. Nguyen, H. Wong, A. Klepchukov, S. Patil, O. Fox and D. Patterson, “CloudStone: Multi-Platform, Multi-Languarge Benchmark and Measurement Tools for Web 2.8,” 2008.
[2] N. Grozev, “Automated CloudStone Setup in Ubuntu VMs / Advanced Automated CloudStone Setup in Ubuntu VMs [Part 2],” 2 June 2014. https://nikolaygrozev.wordpress.com/tag/cloudstone/.
[3] java.net, “Faban Harness and Benchmark Framework,” 11 May 2014. http://java.net/projects/faban/.
[4] S. Lohr, “For Impatient Web Users, an Eye Blink Is Just Too Long to Wait,” 29 February 2012. http://www.nytimes.com/2012/03/01/technology/impatient-web-usersflee-slow-loading-sites.html.
[5] J. Heo, “Deploying Extremely Latency-Sensitive Applications in VMware vSphere 5.5,” 18 September 2013.   http://blogs.vmware.com/performance/2013/09/deploying-extremely-latency-sensitive-applications-in-vmware-vsphere-5-5.html.




Virtual SAN and SAP IQ – a Perfect Match

A performance study shows that VMware vSphere 5.5 with Virtual SAN as the storage backend provides an excellent platform for virtualized deployments of SAP IQ Multiplex Servers.

We created four virtual machines with the RHEL 6.3 operating system, and these virtual machines made up the SAP IQ Multiplex Server, which used Virtual SAN as its storage backend. In order to measure performance, we looked at the distributed query processing (DQP) modes of SAP IQ. In DQP, work is performed by threads running on both leader and worker nodes, and intermediate results are transmitted between these nodes through a shared disk space, or over an inter-node network. In the paper, we refer to these modes as storage-transfer and network-transfer.

In a test consisting of concurrent streams of queries designed to emulate a multi-user scenario, we found that the read-heavy I/O profile of this workload takes full advantage of the Virtual SAN’s flash acceleration layer. Data read from magnetic disks in each disk group, is cached in the SSD in the disk group. Since 70% of SSD capacity is reserved for the read cache, a significant amount of data is quickly placed in very low latency storage. Once it is warmed up, I/O requests are served from the read cache, leading to fast query response times. Add to this SAP IQ’s ability to use network resources to handle intermediate results transfer and we get an additional bump in throughput since we no longer have the overhead of writing intermediate, shared results to disk.

Read more about Distributed Query Processing in SAP IQ on VMware vSphere and Virtual SAN.

Docker Containers Performance in VMware vSphere

By  Qasim Ali,  Banit Agrawal, and Davide Bergamasco


“Containers without compromise” – This was one of the key messages at VMworld 2014 USA in San Francisco. It was presented in the opening keynote, and then the advantages of running Docker containers inside of virtual machines were discussed in detail in several breakout sessions. These include security/isolation guarantees and also the existing rich set of management functionalities. But some may say, “These benefits don’t come for free: what about the performance overhead of running containers in a VM?”

A recent report compared the performance of a Docker container to a KVM VM and showed very poor performance in some micro-benchmarks and real-world use cases: up to 60% degradation. These results were somewhat surprising to those of us accustomed to near-native performance of virtual machines, so we set out to do similar experiments with VMware vSphere. Below, we present our findings of running Docker containers in a vSphere VM and  in a native configuration. Briefly,

  • We find that for most of these micro-benchmarks and Redis tests, vSphere delivered near-native performance with generally less than 5% overhead.
  • Running an application in a Docker container in a vSphere VM has very similar overhead of running containers on a native OS (directly on a physical server).

Next, we present the configuration and benchmark details as well as the performance results.

Deployment Scenarios

We compare four different scenarios as illustrated below:

  • Native: Linux OS running directly on hardware (Ubuntu, CentOS)
  • vSphere VM: Upcoming release of vSphere with the same guest OS as native
  • Native-Docker: Docker version 1.2 running on a native OS
  • VM-Docker: Docker version 1.2 running in guest VM on a vSphere host

In each configuration all the power management features are disabled in the BIOS and Ubuntu OS.

Test Scenarios

Figure 1: Different test scenarios


For this study, we used the micro-benchmarks listed below and also simulated a real-world use case.

–   Micro-benchmarks:

  • LINPACK: This benchmark solves a dense system of linear equations. For large problem sizes it has a large working set and does mostly floating point operations.
  • STREAM: This benchmark measures memory bandwidth across various configurations.
  • FIO: This benchmark is used for I/O benchmarking for block devices and file systems.
  • Netperf: This benchmark is used to measure network performance.

Real-world workload:

  • Redis: In this experiment, many clients perform continuous requests to the Redis server (key-value datastore).

For all of the tests, we run multiple iterations and report the average of multiple runs.

Performance Results


LINPACK solves a dense system of linear equations (Ax=b), measures the amount of time it takes to factor and solve the system of N equations, converts that time into a performance rate, and tests the results for accuracy. We used an optimized version of the LINPACK benchmark binary based on the Intel Math Kernel Library (MKL).

Hardware: 4 socket Intel Xeon E5-4650 2.7GHz with 512GB RAM, 32 total cores, Hyper-Threading disabled
Software: Ubuntu 14.04.1 with Docker 1.2
VM configuration: 32 vCPU VM with 45K and 65K problem sizes


Figure 2: LINPACK performance for different test scenarios

We disabled HT for this run as recommended by the benchmark guidelines to get the best peak performance. For the 45K problem size, the benchmark consumed about 16GB memory. All memory was backed by transparent large pages. For VM results, large pages were used both in the guest (transparent large pages) and at the hypervisor level (default for vSphere hypervisor). There was 1-2% run-to-run variation for the 45K problem size. For 65K size, 33.8GB memory was consumed and there was less than 1% variation.

As shown in Figure 2, there is almost negligible virtualization overhead in the 45K problem size. For a bigger problem size, there is some inherent hardware virtualization overhead due to nested page table walk. This results in the 5% drop in performance observed in the VM case. There is no additional overhead of running the application in a Docker container in a VM compared to running the application directly in the VM.


We used a NUMA-aware  STREAM benchmark, which is the classical STREAM benchmark extended to take advantage of NUMA systems. This benchmark measures the memory bandwidth across four different operations: Copy, Scale, Add, and Triad.

Hardware: 4 socket Intel Xeon E5-4650 2.7GHz with 512GB RAM, 32 total cores, HT enabled
Software: Ubuntu 14.04.1 with Docker 1.2
VM configuration: 64 vCPU VM (Hyper-Threading ON)


Figure 3: STREAM performance for different test scenarios

We used an array size of 2 billion, which used about 45GB of memory. We ran the benchmark with 64 threads both in the native and virtual cases. As shown in Figure 3, the VM added about 2-3% overhead across all four operations. The small 1-2% overhead of using a Docker container on a native platform is probably in the noise margin.


We used Flexible I/O (FIO) tool version 2.1.3 to compare the storage performance for the native and virtual configurations, with Docker containers running in both. We created a 10GB file in a 400GB local SSD drive and used direct I/O for all our tests so that there were no effects of buffer caching inside the OS. We used a 4k I/O size and tested three different I/O profiles: random 100% read, random 100% write, and a mixed case with random 70% read and 30% write. For the 100% random read and write tests, we selected 8 threads and an I/O depth of 16, whereas for the mixed test, we select an I/O depth of 32 and 8 threads. We use the taskset to set the CPU affinity on FIO threads in all configurations. All the details of the experimental setup are given below:

Hardware: 2 socket Intel Xeon E5-2660 2.2GHz with 392GB RAM, 16 total cores, Hyper-Threading enabled
Guest: 32-vCPU  14.04.1 Ubuntu 64-bit server with 256GB RAM, with a separate ext4 disk in the guest (on VMFS5 in vSphere run)
Benchmark:  FIO, Direct I/O, 10GB file
I/O Profile:  4k I/O, Random Read/Write: depth 16, jobs 8, Mixed: depth 32, jobs 8


Figure 4: FIO benchmark performance for different test scenarios

The figure above shows the normalized maximum IOPS achieved for different configurations and different I/O profiles. For random read in a VM, we see that there is about 2% reduction in maximum achievable IOPS when compared to the native case. However, for the random write and mixed tests, we observed almost the same performance (within the noise margin) compared to the native configuration.


Netperf is used to measure throughput and latency of networking operations. All the details of the experimental setup are given below:

Hardware (Server): 4 socket Intel Xeon E5-4650 2.7GHz with 512GB RAM, 32 total cores, Hyper-Threading disabled
Hardware (Client): 2 socket Intel Xeon X5570 2.93GHz with 64GB RAM, 8 cores total, Hyper-Threading disabled
Networking hardware: Broadcom Corporation NetXtreme II BCM57810
Software on server and Client: Ubuntu 14.04.1 with Docker 1.2
VM configuration: 2 vCPU VM with 4GB RAM

The server machine for Native is configured to have only 2 CPUs online for fair comparison with a 2-vCPU VM. The client machine is also configured to have 2 CPUs online to reduce variability. We tested four configurations: directly on the physical hardware (Native), in a Docker container (Native-Docker), in a virtual machine (VM), and in a Docker container inside a VM (VM-Docker). For the two Docker deployment scenarios, we also studied the effect of using host networking as opposed to the Docker bridge mode (default operating mode), resulting in two additional configurations (Native-Docker-HostNet and VM-Docker-HostNet) making total six configurations.

We used TCP_STREAM and TCP_RR tests to measure the throughput and round-trip network latency between the server machine and the client machine using a direct 10Gbps Ethernet link between two NICs. We used standard network tuning like TCP window scaling and setting socket buffer sizes for the throughput tests.


Figure 5: Netperf Recieve performance for different test scenarios


Figure 6: Netperf transmit performance for different test scenarios

Figure 5 and Figure 6 shows the unidirectional throughput over a single TCP connection with standard 1500 byte MTU for both transmit and receive TCP_STREAM cases (We used multiple Streams in VM-Docker* transmit case to reduce the variability in runs due to Docker bridge overhead and get predictable results). Throughput numbers for all configurations are identical and equal to the maximum possible 9.40Gbps on a 10GbE NIC.


Figure 7: Netperf TCP_RR performance for different test scenarios (Lower is better)

For the latency tests, we used the latency sensitivity feature introduced in vSphere5.5 and applied the best practices for tuning latency in a VM as mentioned in this white paper. As shown in Figure 7, latency in a VM with VMXNET3 device is only 15 microseconds more than in the native case because of the hypervisor networking stack. If users wish to reduce the latency even further for extremely latency- sensitive workloads, pass-through mode or SR-IOV can be configured to allow the guest VM to bypass the hypervisor network stack. This configuration can achieve similar round-trip latency to native, as shown in Figure 8. The Native-Docker and VM-Docker configuration adds about 9-10 microseconds of overhead due to the Docker bridge NAT function. A Docker container (running natively or in a VM) when configured to use host networking achieves similar latencies compared to the latencies observed when not running the workload in a container (native or a VM).


Figure 8: Netperf TCP_RR performance for different test scenarios (VMs in pass-through mode)


We also wanted to take a look at how Docker in a virtualized environment performs with real world applications. We chose Redis because: (1) it is a very popular application in the Docker space (based on the number of pulls of the Redis image from the official Docker registry); and (2) it is very demanding on several subsystems at once (CPU, memory, network), which makes it very effective as a whole system benchmark.

Our test-bed comprised two hosts connected by a 10GbE network. One of the hosts ran the Redis server in different configurations as mentioned in the netperf section. The other host ran the standard Redis benchmark program, redis-benchmark, in a VM.

The details about the hardware and software used in the experiments are the following:

Hardware: HP ProLiant DL380e Gen8 2 socket Intel Xeon E5-2470 2.3GHz with 96GB RAM, 16 total cores, Hyper-Threading enabled
Guest OS: CentOS 7
VM: 16 vCPU, 93GB RAM
Application: Redis 2.8.13
Benchmark: redis-benchmark, 1000 clients, pipeline: 1 request, operations: SET 1 Byte
Software configuration: Redis thread pinned to CPU 0 and network interrupts pinned to CPU 1

Since Redis is a single-threaded application, we decided to pin it to one of the CPUs and pin the network interrupts to an adjacent CPU in order to maximize cache locality and avoid cross-NUMA node memory access.  The workload we used consists of 1000 clients with a pipeline of 1 outstanding request setting a 1 byte value with a randomly generated key in a space of 100 billion keys.  This workload is highly stressful to the system resources because: (1) every operation results in a memory allocation; (2) the payload size is as small as it gets, resulting in very large number of small network packets; (3) as a consequence of (2), the frequency of operations is extremely high, resulting in complete saturation of the CPU running Redis and a high load on the CPU handling the network interrupts.

We ran five experiments for each of the above-mentioned configurations, and we measured the average throughput (operations per second) achieved during each run.  The results of these experiments are summarized in the following chart.


Figure 9: Redis performance for different test scenarios

The results are reported as a ratio with respect to native of the mean throughput over the 5 runs (error bars show the range of variability over those runs).

Redis running in a VM has slightly lower performance than on a native OS because of the network virtualization overhead introduced by the hypervisor. When Redis is run in a Docker container on native, the throughput is significantly lower than native because of the overhead introduced by the Docker bridge NAT function. In the VM-Docker case, the performance drop compared to the Native-Docker case is almost exactly the same small amount as in the VM-Native comparison, again because of the network virtualization overhead.  However, when Docker runs using host networking instead of its own internal bridge, near-native performance is observed for both the Docker on native hardware and Docker in VM cases, reaching 98% and 96% of the maximum throughput respectively.

Based on the above results, we can conclude that virtualization introduces only a 2% to 4% performance penalty.  This makes it possible to run applications like Redis in a Docker container inside a VM and retain all the virtualization advantages (security and performance isolation, management infrastructure, and more) while paying only a small price in terms of performance.


In this blog, we showed that in addition to the well-known security, isolation, and manageability advantages of virtualization, running an application in a Docker container in a vSphere VM adds very little performance overhead compared to running the application in a Docker container on a native OS. Furthermore, we found that a container in a VM delivers near native performance for Redis and most of the micro-benchmark tests we ran.

In this post, we focused on the performance of running a single instance of an application in a container, VM, or native OS. We are currently exploring scale-out applications and the performance implications of deploying them on various combinations of containers, VMs, and native operating systems.  The results will be covered in the next installment of this series. Stay tuned!


Custom Power Management Settings for Power Savings in vSphere 5.5

VMware vSphere serves as a common virtualization platform for a diverse ecosystem of applications. Every application has different performance demands which must be met, but the power and cooling costs of running these applications are also a concern. vSphere’s default power management policy, “Balanced”, meets both of these goals by effectively preserving system performance while still saving some power.

For those who would like to prioritize energy efficiency even further, vSphere provides additional ways to tweak its power management under the covers. Custom power management settings in ESXi let you create your own power management policy, and your server’s BIOS also typically lets you customize hardware settings which can maximize power savings at a potential cost to performance.

When choosing a low power setting, we need to know whether it is effective at increasing energy efficiency, that is, the amount of work achieved for the power consumed. We also need to know how large of an impact the setting has on application throughput and latencies. A power saving setting that is too aggressive can result in low system performance. The best combination of power saving techniques will be highly individualized to your workload; here, we present one case study.

We used the VMmark virtualization benchmark to measure the effect of ESXi custom power settings and BIOS custom settings on energy efficiency and performance. VMmark 2.5 is a multi-host virtualization benchmark that uses diverse application workloads as well as common platform level workloads to model the demands of the datacenter. VMs running a complete set of the application workloads are grouped into units of load called tiles. For more details, see the VMmark 2.5 overview.

In this study, the best custom power setting produced an increase in energy efficiency of 17% with no significant drop in performance at moderate levels of load.

Test Methodology

All tests were conducted on a two-node cluster running VMware vSphere 5.5 U1. Each custom power management setting was tested independently to gauge its effects on energy efficiency and performance while all other settings were left at their defaults. The settings tested fall into two categories: ESXi custom power settings and BIOS custom settings. We discuss how to modify these settings at the end of the article.

Systems Under Test: Two Dell PowerEdge R720 servers
Configuration Per Server  
            CPUs: Two 12-core Intel® Xeon® E5-2697 v2 @ 2.7 GHz, Turbo Boost Enabled, up to 3.5 GHz, Hyper-Threading enabled
            Memory: 256GB ECC DDR3 @ 1866MHz
            Host Bus Adapter: QLogic ISP2532 Dual Port 8Gb Fibre Channel to PCI Express
           Network Controller: Integrated Intel I350 Quad-Port Gigabit Adapter, one Intel I350 Dual-Port Gigabit PCIe Adapter
            Hypervisor: VMware ESXi 5.5 U1
Shared Resources  
            Virtualization Management: VMware vCenter Server 5.5
            Storage Array: EMC VNX5800
30 Enterprise Flash Drives (SSDs) and 32 HDDs, grouped as two 10-SSD RAID0 LUNs and four 8-HDD RAID0 LUNs. FAST Cache was configured from 10 SSDs.
            Power Meters: One Yokogawa WT210 per server

Each configuration was tested at five different load points: 1 tile (the lowest load level), 4, 7, 10, and 12 tiles, which was the maximum number of tiles that met Quality of Service (QoS) requirements. All datapoints are the mean of three tests in each configuration.

ESXi Custom Power Settings

ESXi custom power settings influence the power state of the processor. We tested two custom power management settings which had the greatest impact on our workload: Power.MaxFreqPct and Power.CstateResidencyCoef. The advanced ESXi setting Power.MaxFreqPct (default value 100) reduces the processor frequency by placing a cap on the highest operating frequency it can reach. In practice, the processor can operate only at certain set frequencies (P-states), so if the frequency cap requested by ESXi (e.g. 2160MHz) does not match to a set frequency state, the processor will run at the nearest lower frequency state (e.g. 2100MHz). Setting Power.MaxFreqPct = 99 put the cap at 99% of the processor’s nominal frequency, which limited Turbo Boost. Power.MaxFreqPct = 80 further limited the maximum frequency of the processor to 80% of its nominal frequency of 2.7GHz, for a maximum of 2.1GHz. Setting Power.CstateResidencyCoef = 0 (default value 5) puts the processor into its deepest available C-state, or lowest power state, when it is idle. As a prerequisite, deep C-states must be enabled in the BIOS. For a more in-depth discussion of power management techniques and other custom options, please see the vSphere documentation and the whitepaper Host Power Management in VMware vSphere 5.5.

VMmark models energy efficiency as performance score per kilowatt of power consumed. VMmark scores in the graph below have been normalized to the default “Balanced” 1-tile result, which does not use any custom power settings.

VMware ESXi Custom Power Management Settings improve efficiency

A major trend can be seen here; an increase in load is correlated with greater energy efficiency. As the CPUs become busier, throughput increases at a faster rate than the required power. This can be understood by noting that an idle server will still consume power, but with no work to show for it. A highly utilized server is typically the most energy efficient per request completed, and the results bear this out.

To more closely examine the relative impact of each custom setting compared to the default setting, we normalized all results within each load level to the default “Balanced” result for that number of tiles. The figure below shows the percent change at each load level.

VMware ESXi Custom Power Management Settings Change in Efficiency and Performance Results

All custom settings showed improvements in efficiency compared to the default “Balanced” setting. The improvements varied depending on load. Setting MaxFreqPct to 99 had the greatest benefit to energy efficiency, between 5% and 15% at varying load levels. The greatest improvement was seen at 4 tiles, which increased efficiency by 17%, while resulting in a performance decrease of only 3%. The performance cost increased with load to 9% at 12 tiles. However, limiting processor frequency even further to a maximum of 80% of its nominal frequency does not produce an additive effect. Not only did efficiency actually decrease relative to MaxFreqPct=99, but it profoundly curtailed performance from 96% of baseline at light load to 84% of baseline for a heavily loaded machine. CstateResidency=0 produced some modest increases in efficiency for a lightly loaded server, but the effect disappeared at higher load levels.

VMmark 2.5 performance scores are based on application and infrastructure workload throughput, while application latency reflects Quality of Service. For the Mail Server, Olio, and DVD Store 2 workloads, latency is defined as the application’s response time. We wanted to see how custom power management settings affected application latency as opposed to the VMmark score. All latencies are normalized to the lowest 1-tile results.

VMware ESXi Custom Power Management Settings Effect on Application Latencies

Naturally, latencies increase as load increases from 1 to 12 tiles. Fortunately, the custom power management policies caused only minimal increases in application latencies, if any, except for the MaxFreqPct=80 setting which did create elevated latencies across the board.

BIOS Custom Power Settings

The Dell PowerEdge R720 BIOS provides another toolbox of power-saving knobs to tweak. Using the BIOS settings, we manually disabled Turbo Boost and reduced memory frequency from its default maximum speed of 1866MT/s (megatransfers per second) to either 1333MT/s or 800MT/s.


The Turbo Boost Disabled configuration produced the largest increase in efficiency, while 800MT/s memory frequency actually decreased efficiency at the higher load levels.
Again, we normalized all results within each load level to its default “Balanced” result. The figure below shows the percent change at each load level.

Disabling Turbo Boost was the most effective setting to increase energy efficiency, with a performance cost of 2% at low load levels to 8% at high load levels. Reducing memory frequency to 1333MT/s had a reliable but small boost to efficiency and no effect on performance, leading us to conclude that a memory speed of 1866MT/s is simply faster than needed for this workload.

Disabling Turbo Boost and reducing memory frequency to 800MT/s increased DVD Store 2 latencies at 10 tiles by 10% and 12 tiles by 30%, but all latencies were still well within Quality of Service requirements.  Reducing memory frequency to 1333MT/s had no effect on application latencies.

Reducing the use of Turbo Boost, using either ESXi custom setting MaxFreqPct or BIOS custom settings, proved to be the most effective way to increase energy efficiency in our VMmark tests. The impact on performance was small, but increased with load. MaxFreqPct is the preferred setting because, like all ESXi custom power management settings, it takes effect immediately and can easily be reversed without reboots or downtime. Other custom power management settings produced modest gains in efficiency, but, if taken to the extreme, not only harm performance but fail to increase efficiency. In addition, energy efficiency is strongly related to load; the most efficient server is also one that is heavily utilized. Taking steps to increase server utilization, such as server consolidation, is an important part of a power saving strategy. Custom power management settings can produce gains in energy efficiency at a cost to performance, so consider the tradeoff when choosing custom power management settings for your own environment.

 How to Configure Custom Power Management Settings

Disclaimer: The results presented above are a case study of the impact of custom power management settings and a starting point only. Results may not apply to your environment and do not represent best practices.

Exercise caution when choosing a custom power management setting. Change settings one at a time to evaluate their impact on your environment. Monitor your server’s power consumption either through its UPS, or consult your vendor to find the rated accuracy of your server’s internal power monitoring sensor. If it is highly accurate, you can view the server’s power consumption in esxtop (press ‘p’ to view Power Usage).

To customize power management settings, enter your server’s BIOS. Power Management settings vary by vendor but most include “OS Controlled” and “Custom” policies.

In the Dell PowerEdge R720, choosing the “Performance Per Watt (OS)” System Profile allows ESXi to control power management, while leaving hardware settings at their default values.

Screenshot of R720 BIOS Selecting OS controlled power managment

Choosing the “Custom” System Profile and setting CPU Power Management to “OS DBPM” allows ESXi to control power management while enabling custom hardware settings.


Using ESXi Custom Power Settings

To enable the vSphere custom power management policy,

  1. Browse to the host in the vSphere Web Client navigator.
  2. Click the Manage tab and click Settings.
  3. Under Hardware, select Power Management and click the Edit button.
  4. Select the Custom power management policy and click OK.

The power management policy changes immediately and does not require a server reboot.


To modify ESXi custom power management settings,

  1. Browse to the host in the vSphere Web Client navigator.
  2. Click the Manage tab and click Settings.
  3. Under System, select Advanced System Settings.
  4. Power management parameters that affect the Custom policy have descriptions that begin with In Custom policy. All other power parameters affect all power management policies.
  5. Select the parameter and click the Edit button.

Note: The default values of power management parameters match the Balanced policy.



Virtual SAP HANA Achieves Production Level Performance

VMware CEO Pat Gelsinger announced production support for SAP HANA on VMware vSphere 5.5 at EMC World this week during his keynote. This is the end result of a very thorough joint testing project over the past year between VMware and SAP.

HANA is an in-memory platform (including database capabilities) from SAP that has enabled huge gains in performance for customers and has been a high priority for SAP over the past few years.  In order for HANA to be supported in a virtual machine on vSphere 5.5 for production workloads, we worked closely with SAP to enable, design, and measure in-depth performance tests.

In order to enable the testing and ongoing production support of SAP HANA on vSphere, two HANA appliance servers were ordered, shipped, and installed into SAP’s labs in Waldorf Germany.  These systems are dedicated to running SAP HANA on vSphere onsite at SAP.  Each system is an Intel Xeon E7-8870 (Westmere-EX) based four-socket server with 1TB of RAM.  They are used for performance testing and also for ongoing support of HANA on vSphere.  Additionally, VMware has onsite support engineering to assist with the testing and support.

SAP designed an extensive performance test suite that used a large number of test scenarios to stress all functions and capabilities of HANA running on vSphere 5.5.  They included OLAP and OLTP with a wide range of data sizes and query functions. In all, over one thousand individual test cases were used in this comprehensive test suite.  These same tests were run on identical native HANA systems and the difference between native and virtual tests was used as the key performance indicator.

In addition, we also tested vSphere features including vMotion, DRS, and VMware HA with virtual machines running HANA.  These tests were done with the HANA virtual machine under heavy stress.

The test results have been extremely positive and are one of the key factors in the announcement of production support.  The difference between virtual and native HANA across all the performance tests was on average within a few percentage points.

The vMotion, DRS, and VMware HA tests were all completed without issues.  Even with the large memory sizes of HANA virtual machines, we were still able to successfully migrate them with vMotion while under load with no issues.

One of the results of the extensive testing is a best practices guide for HANA on vSphere 5.5. This document includes a performance guide for running HANA on vSphere 5.5 based on this extensive testing.  The document also includes information about how to size a virtual HANA instance and how VMware HA can be used in conjunction with HANA’s own replication technology for high availability.

Power Management and Performance in VMware vSphere 5.1 and 5.5

Power consumption is an important part of the datacenter cost strategy. Physical servers frequently offer a power management scheme that puts processors into low power states when not fully utilized, and VMware vSphere also offers power management techniques. A recent technical white paper describes the testing and results of two performance studies: The first shows how power management in VMware vSphere 5.5 in balanced mode (the default) performs 18% better than the physical host’s balanced mode power management setting. The second study compares vSphere 5.1 performance and power savings in two server models that have different generations of processors. Results show the newer servers have 120% greater performance and 24% improved energy efficiency over the previous generation.

For more information, please read the paper: Power Management and Performance in VMware vSphere 5.1 and 5.5.

Line-Rate Performance with 80GbE and vSphere 5.5

With the increasing number of physical cores in a system, the networking bandwidth requirement per server has also increased. We often find many networking-intensive applications are now being placed on a single server, which results in a single vSphere server requiring more than one 10 Gigabit Ethernet (GbE) adapter. Additional network interface cards (NICs) are also deployed to separate management traffic and the actual virtual machine traffic. It is important for these servers to service the connected NICs well and to drive line rate on all the physical adapters simultaneously.

vSphere 5.5 supports eight 10GbE NICs on a single host, and we demonstrate that a host running with vSphere 5.5 can not only drive line rate on all the physical NICs connected to the system, but can do it with a modest increase in overall CPU cost as we add more NICs.

We configured a single host with four dual-port Intel 10GbE adapters for the experiment and connected them back-to-back with an IXIA Application Network Processor Server with eight 10GbE ports to generate traffic. We then measured the send/receive throughput and the corresponding CPU usage of the vSphere host as we increased the number of NICs under test on the system.

Environment Configuration

  • System Under Test: Dell PowerEdge R820
  • CPUs: 4 x  Intel Xeon Processors E5-4650 @ 2.70GHz
  • Memory: 128GB
  • NICs:8 x Intel 82599EB 10GbE, SFP+ Network Connection
  • Client: Ixia Xcellon-Ultra XT80-V2, 2U Application Network Processor Server

Challenges in Getting 80Gbps Throughput

To drive near 80 gigabits of data per second from a single vSphere host, we used a server that has not only the required CPU and memory resources, but also the PCI bandwidth that can perform the necessary I/O operations. We used a Dell PowerEdge Server with an Intel E5-4650 processor because it belongs to the first generation of Intel processors that supports PCI Gen 3.0. PCI Gen 3.0 doubles the PCI bandwidth capabilities compared to PCI Gen 2.0. Each dual-port Intel 10GbE adapter needs at least a PCI Gen 2.0 x8 to reach line rate. Also, the processor has Intel Data Direct I/O Technology where the packets are placed directly in the processor cache rather than going to the memory. This reduces the memory bandwidth consumption and also helps reduce latency.

Experiment Overview

Each 10GbE port of the vSphere 5.5 server was configured with a separate vSwitch, and each vSwitch had two Red Hat 6.0 Linux virtual machines running an instance of Apache web server. The web server virtual machines were configured with 1 vCPU and 2GB of memory with VMXNET3 as the virtual NIC adapter.  The 10GbE ports were then connected to the Ixia Application Server port. Since the server had two x16 slots and five x8 slots, we used the x8 slots for the four 10GbE NICs so that each physical NIC had identical resources. For each physical connection, we then configured 200 web/HTTP connections, 100 for each web server, on an Ixia server that requested or posted the file. We used a high number of connections so that we had enough networking traffic to keep the physical NIC at 100% utilization.

Figure 1. System design of NICs, switches, and VMs

The Ixia Xcellon application server used an HTTP GET request to generate a send workload for the vSphere host. Each connection requested a 1MB file from the HTTP web server.

Figure 2 shows that we could consistently get the available[1] line rate for each physical NIC as we added more NICs to the test. Each physical NIC was transmitting 120K packets per second and the average TSO packet size was close to 10K. The NIC was also receiving 400K packets per second for acknowledgements on the receive side. The total number of packets processed per second was close to 500K for each physical connection.

Figure 2. vSphere 5.5 drives throughput at available line rates. TSO on the NIC resulted in lower packets per second for send.

Similar to the send case, we configured the application server to post a 1MB file using an HTTP POST request for generating receive traffic for the vSphere host. We used the same number of connections and observed similar behavior for the throughput. Since the NIC does not have support for hardware LRO, we were getting 800K packets per second for each NIC. With eight 10GbE NICs, the packet rate reached close to 6.4 million packets per second. VMware does Software LRO for Linux and as a result we see large packets in the guest. The guest packet rate is around 240K packets per second. There was also significant traffic for TCP acknowledgements and for each physical NIC. The host was transmitting close to 120K acknowledgement packets for each physical NIC, bringing the total packets processed close to 7.5 million packets per second for eight 10Gb ports.

Figure 3. Average vSphere 5.5 host CPU utilization for send and receive

We also measured the average CPU reported for each of the tests. Figure 3 shows that the vSphere host’s CPU usage increased linearly as we added more physical NICs to the test for both send and receive. This indicates that performance improves at an expected and acceptable rate.

Test results show that vSphere 5.5 is an excellent platform on which to deploy networking-intensive workloads. vSphere 5.5 makes use of all the physical bandwidth capacity available and does this without incurring additional CPU cost.


[1]A 10GbE NIC can achieve only 9.4 Gbps of throughput with standard MTU. For a 1500 byte packet, we have 40 bytes for the TCP /IP header and 38 bytes for the Ethernet frame format.

Deploying Extremely Latency-Sensitive Applications in VMware vSphere 5.5

VMware vSphere ensures that virtualization overhead is minimized so that it is not noticeable for a wide range of applications including most business critical applications such as database systems, Web applications, and messaging systems. vSphere also supports well applications with millisecond-level latency constraints, including VoIP services. However, performance demands of latency-sensitive applications with very low latency requirements such as distributed in-memory data management, stock trading, and high-performance computing have long been thought to be incompatible with virtualization.

vSphere 5.5 includes a new feature for setting latency sensitivity in order to support virtual machines with strict latency requirements. This per-VM feature allows virtual machines to exclusively own physical cores, thus avoiding overhead related to CPU scheduling and contention. A recent performance study shows that using this feature combined with pass-through mechanisms such as SR-IOV and DirectPath I/O helps to achieve near-native performance in terms of both response time and jitter.

The paper explains major sources of latency increase due to virtualization in vSphere and presents details of how the latency-sensitivity feature improves performance along with evaluation results of the feature. It also presents some best practices that were concluded from the performance evaluation.

For more information, please read the full paper: Deploying Extremely Latency-Sensitive Applications in VMware vSphere 5.5.


Exploring Generational Differences in Performance and Energy Efficiency Using VMware VMmark 2.5

Each new generation of servers brings advances in hardware components. For IT professionals purchasing or managing new generations of hardware, it’s vital to understand how these incremental hardware improvements translate into real-world gains in the datacenter. Using the VMware VMmark 2.5 virtualization benchmark, we compared performance and energy efficiency of two different generations of servers in four-node clusters.

VMmark 2.5 is a multi-host virtualization benchmark that uses varied application workloads as well as common datacenter operations to model the demands of the datacenter. VMs running diverse application workloads are grouped into units of load called tiles. For more details, see the VMmark 2.5 overview.

Testing Methodology
All tests were conducted on two four-node clusters running VMware vSphere 5.1. We compared performance and energy efficiency between a cluster of previous generation Dell R310 servers, and a cluster of current generation Dell R620 servers. For simplicity, we refer to these as the ‘old cluster’ and ‘new cluster,’ respectively. Among other hardware differences, the old cluster servers contained four-core Intel Nehalem processors while the new cluster servers contained eight-core Intel Sandy Bridge EP processors. Memory in the newer servers was appropriately scaled up to accommodate their increased processing power and represents common current server configurations. Software and storage configurations were identical between clusters.

Old Cluster
Systems Under Test: Four Dell PowerEdge R310 servers
CPUs (per server): One Quad-Core Intel® Xeon® X3460 @ 2.8 GHz, Hyper-Threading enabled
Memory (per server): 32GB DDR3 ECC @ 800 MHz

New Cluster
Systems Under Test: Four Dell PowerEdge R620 servers
CPUs (per server): One Eight-Core Intel® Xeon® E5-2665 @ 2.4 GHz, Hyper-Threading enabled
Memory (per server): 96GB DDR3 ECC @ 1067 MHz

Storage Array: EMC VNX5700
        62 Enterprise Flash Drives (SSDs), RAID 0, grouped as 3 x 8 SSD LUNs, 7 x 5 SSD LUNs, and 1 x 3 SSD LUN
Hypervisor: VMware vSphere 5.1.0
Virtualization Management: VMware vCenter Server 5.1.0
VMmark version: 2.5

To determine the maximum VMmark load the old cluster could support, we increased the number of VMmark tiles until the cluster reached saturation, which is defined as the largest number of tiles that still meet Quality of Service (QoS) requirements. We then tested the new cluster at the same number of tiles. All data points are the mean of four tests in each configuration and VMmark scores are normalized to the old cluster’s performance.

The new cluster had a 32% higher VMmark score in combination with a 41% lower CPU utilization. The new cluster also showed a 24% increase in energy efficiency over the old cluster, which we’ll discuss further below. At four tiles, the old cluster was bottlenecked on CPU, resulting in decreased workload throughput, while the new cluster was not. With CPU resources to spare, the new cluster met the requested load at lower latencies, which increased its total throughput and score. Mean I/O latencies remained low for both clusters at 1.2ms reads and 1.1ms writes for the old cluster and 1.0ms reads and 0.9ms writes for the new cluster.

We next determined the maximum VMmark load the new cluster could support. While the old cluster was saturated at four tiles, the new cluster accommodated more than twice the load at nine tiles and produced a score 120% higher than the old cluster. Mean I/O latencies remained low at 1.0ms.

Click to enlarge

The performance advantages of the R620 over the R310 were largely due to the generational improvements of the R620’s eight-core E5-2665 processor versus the R310’s four-core x3460 processor, which includes improved bus speeds and larger L3 cache, and the R620’s increased memory.

These performance results suggest that it would be possible to replace four Dell R310 servers with two Dell R620 servers and expect better than equivalent performance. We put this to the test by removing two nodes from the new cluster and found that the two remaining nodes did support four tiles at 93% utilization, with an 11% higher VMmark score and 74% greater energy efficiency than the four-host old cluster.

Beyond their raw performance capability, we also compared the two server generations on their energy efficiency. The Performance per Kilowatt metric, which is new to VMmark 2.5, models energy efficiency as VMmark score per kilowatt of power consumed. Below, we’ve plotted energy efficiency against the normalized VMmark score. Both clusters were run with their servers’ power management set to “maximum performance.”

Energy Efficiency as a Function of VMmark 2.5 Score

Two trends emerge from this figure. First, at four tiles, the four-host new cluster accomplishes more work at higher energy efficiency than the old cluster. Across the board, the new cluster is more energy efficient than the old cluster. Second, within the four-host new cluster, greater energy efficiency is correlated with increase in VMmark score. As the CPUs become busier, performance increases at a faster rate than the required power. This can be understood by noting that an idle server will still consume power, but with no performance to show for it. A highly utilized server is typically the most energy efficient per request completed, which is confirmed by the two-host new cluster that achieved high efficiency at 93% utilization. Higher energy efficiency creates cost savings in energy consumption and in cooling costs.

Our investigation shows that, while running vSphere 5.1, two newer Dell R620 servers are capable of supporting a greater load than four older Dell R310 servers. Because the Dell R620 performance is more than double that of the Dell R310, a four-node Dell R620 cluster reached a 120% higher maximum score than the Dell R310 cluster. In addition to its performance advantages, at each load level the Dell R620 cluster performed with greater energy efficiency, showing that the Dell R620 has superior performance but also has greater energy efficiency than the Dell R310.