Emergence of new technology often brings with it new solutions that take advantage of the capabilities in interesting ways to solve problems for businesses. I've recently been speaking with ClearSky and Xtium about an emerging model for storage they are bringing to the market.
With Virtual Volumes as a fundamental building block of their solution, it is my pleasure to provide a guest-authored post from two leaders of these companies, Laz Vekiarides and Tim Vogel. Here they will expand on how their offering helps companies achieve all the benefits of per-VM control and management while preserving performance and availability coupled with the resilience, scale and economics of cloud storage.
It's been several years since the first virtual volumes (vVols) demonstrations at VMworld 2012, and it’s extremely exciting to see the technology go live this year. To us in the storage industry, vVols represented the most elegant solution to one of the biggest administrative problems that exist in virtual infrastructure: the mapping between virtual machines and their underlying storage. Up until now, storage management in virtual infrastructure has been very LUN-centric – particularly for things like snapshots, clones, and data protection. To an administrator who is supporting workloads and VM’s, this adds another layer of unwanted complexity. Rather than being able to focus on the needs of the individual workloads, they have to worry about the characteristics of this shared storage resource. Is it overloaded? Is a single “bad” VM making life worse for everything else? Is there enough free space? How to split things up if another LUN is added? In my previous life, I would spend untold hours and resources helping customers deal with these intractable issues. In fact, some of the biggest barriers to building a true self-service private cloud experience have centered around these storage problems.
vVols Storage changes the game in two very important ways. First, it allows the administration process to be VM-centric, the way it is supposed to be. Being able to manage storage one virtual workload at a time allows enterprise IT to more tightly control the performance of each workload and more accurately map it to the right resources. More important, vVols frees the VI administrator from having to worry about the vagaries of provisioning storage – adding a new application can be a point and click operation. Worrying about LUN capacities and your VMs’ neighbors becomes a thing of the past because every single VM has its very own unique storage resources.
Second, and perhaps more relevant for service providers, vVols allows truly virtualized storage infrastructure to shine. The first generation of virtualized storage arrays were designed around the requirements of the physical datacenter. We built them to surface LUNs and behave, more or less, like their physical counterparts. Hence, the numbers of LUNs, usage models, and general system scale was the same order of magnitude as the physical arrays they replaced. This was fine, because there was no use case that required them to behave otherwise. As we transition fully into the Cloud Era, we now have that use case, and we have a whole range of new cloud technologies that can truly change the way shared storage is consumed. The time is ripe to rethink storage for today’s hybrid cloud environment.
The ClearSky Global Storage Network:
Built for vVols Storage from the ground up
The ClearSky managed service is comprised of a network of caching resources residing in metro areas across the world. Customers can connect to this edge at their local datacenters using high speed Ethernet connectivity without having to stand up a large footprint of hardware. The ClearSky service is backed entirely by scale-out object storage clouds from leading service providers. This allows for nearly infinite scalability and permits customers of the ClearSky service to allocate resources on demand without any pre-provisioning. This operationally infinite resource also provides a number of additional technical benefits that enable enterprise class storage management in a completely new way.
Unlike legacy, physical storage systems, the ClearSky global storage network was designed from the beginning to support virtual volumes and per-workload granularity. Built on our own state of the art storage virtualization technology, the ClearSky infrastructure has all of the advanced features that VMware administrators and users of enterprise storage would expect, without the physical limitations that they have become accustomed to. Most important, because storage capacity is accessed via a private network through an edge endpoint, users no longer have any of the management and physical data protection headaches attendant with physical array infrastructure. Just plug in!
Because ClearSky is backed by the vast, highly durable object storage of the cloud, data capacity is effectively unconstrained. This allows the service to have enormous scalability, without the limits of traditional systems. Hundreds of thousands of vVols can be supported with very deep snapshot trees and with no performance degradation from snapshots. With built-in data optimization and encryption technology and full integration into vSphere, the ClearSky service offloads enormous amounts of storage-related computation to the network. This allows greater VM density with existing servers and leaves more computational power available to workloads.
Fully managed data protection
Rather than protecting data via RAID at the compute node, the system takes advantage of the cloud’s extraordinary durability while shielding customers from the high latency inherent to the regional nature of clouds. Full copies of all data reside in the cloud while the system maintains warm and hot cache tiers in a network of metro-based caches that sit close to where compute and users live. With all of the data management functionality residing in the ClearSky network, customers no longer need to hassle with configuring RAID, backup, disaster recovery, and associated networking and infrastructure in order to maintain a high level of availability and redundancy for their users’ data. Highly available, highly durable, and highly performant data storage resources can be accessed simply by plugging in.
True location independence for hybrid scenarios
Because the backing storage for the network lives in scale-out clouds, it is now possible to enable true hybrid operational scenarios. Data does not need to be moved back and forth from datacenter to cloud. The same storage resources visible at the datacenter can be surfaced in a compute cloud. Thus moving a workload for disaster recovery or even for convenience can simply be done by spinning it down in at the datacenter and spinning it up in the cloud without migration of massive amounts of data.
Metro-based migration and failover
As a metro area caching system, additional use cases that were previously achievable only at great expense are now possible. Taking advantage of the proximity of hot and warm data within a metro area, customers can enable migration of workloads across datacenter locations. Metro area clusters can be implemented simply by standing up compute resources and connecting to the network. Workloads can move across locations with a simple series of clicks.
A Platform for Service Providers
Today, legacy technologies make it challenging and costly for service providers to manage compute hosting and DR services. The only way to do this is to set up additional islands of expensive storage, make additional copies of customer data using replication, and then deal with security issues and complex failover and failback scenarios to manage the versioning of the data. Given the complexity and expense, it is not feasible to do this in an on-demand, agile fashion, and it is limited to a small number of geographies.
With location independence, ClearSky’s global storage network creates a new platform for service providers to offer value-added services enabled by the advanced features of the network. Secure, high performance access to data can be available across the network to any metro area where the service is offered. Additionally, if compute resources aren’t available on-premise, connectivity adjacent to public cloud resources such as vCloud Air will allow the option of leveraging pay-as-you-go compute on a per VM basis using vVols.
ClearSky has partnered with Xtium, a cloud and managed service provider based in Philadelphia, to provide cloud services to joint customers by leveraging the ClearSky service and vVols Storage. Xtium delivers its services to customers across several U.S. locations and also through vCloud Air. With ClearSky’s global storage network, Xtium has eliminated the costs and management of replicating data in sync at multiple backup sites, while increasing the compute available at those sites to provide additional value added services. With the ClearSky vVols solution, not only has Xtium has been able to greatly simplify the management of their customers’ workloads, but has added new capabilities such as data mobility as well.
Visit the ClearSky Data team in booth #441 at VMworld 2015 to learn more about leveraging vVOLs with the ClearSky service, and attend the VMware-led session, “What’s New in Virtual Volumes” (STO5571), with guests Laz Vekiarides, ClearSky Data chief technology officer, and Tim Vogel, chief technology officer of ClearSky Data customer Xtium, on August 31 from 2 to 3 p.m.