Observations from a VMworld Walkabout

Walk along the GlacierTook several walks around the show floor this week at VMworld and noticed a ton of flash storage vendors along with established companies showing off their latest SSD based arrays.  We’re in very exciting times for VMware administrators looking to run their production apps in a virtual or cloud environment.  It was only yesterday that Oracle DBA’s and Microsoft admins would never have dared to virtualize their tier-1 mission critical apps.  This can be attributed to 3 fundamental changes over the past 5 years:

Trust.  VMware has been steadily enhancing the ESX, VI2, VI3, vSphere platform to have the resources to run these apps.  As VMware’s platform’s gotten more performant, starting small with print/file servers and moving up to the SAP, Oracle, Exchange big boys club, organizations have learned to trust virtualization.  Most importantly, the ability to support these apps in the 24/7, always-on world with fewer resources has been a life-saver for many organizations.

Support. It was only in 2007 that if you’d dared to virtualized MS Exchange or Oracle you’d have an ‘Unsupported Configuration’ on your hands. That’s when it was VMware’s word against the application vendor, with you, the end user stuck in the middle.  Not a happy place for a virtualization administrator trying to help his or her organization.  Now it’s a distant memory with most tier-1 apps supported on VMware and/or the app owners flavor of virtualization like Hyper-v/Oracle VM.

Offload. Before offloads, your production ESX server had to be a pricey workhorse of the datacenter, expected to do everything.  With the introduction of API’s such as vStorage, your SAN was able to share more of the load and could perform performance hungry operations like snapshotting.  This ultimately made your VMware environment faster and more scalable.  Copy functions like, snapshots, cloning, replication for data protection, disaster recovery, VDI, and Test/Dev are now offloaded through self-aware pools of compute and production/networked storage on almost every popular production array.

So what happens to the copy services in a flashy world?

If our VMware environments can now offload copying operations to production SANs, and those SANs are very expensive, highly performant SSD based systems running our tier-1 apps – Where do we put the copies.  IDC estimates that we’re storing between 13-120 duplicate copies of data across our enterprises. Storing them them on your shiny new Flash array simply isn’t economically viable or feasible.  Yet the systems around our applications, inside and outside our production SANs make copies all day long.  We make copies for all sorts of protection and availability reasons. These copies help with the trust factor of virtualizing tier-1 apps, that in a disaster scenario, we’ll be able to meet our application SLA’s.

From these walkabouts, we believe that two types of storage are emerging:  1. Production Data Storage that will be SSD based 2. Copy Data Storage that will be capacity optimized and SLA driven. With Copy Data Storage you can offload all copy functions from your production server and storage environment, effectively optimizing the entire data center.  This bifurcation of the storage market is happening faster than we realize, in-part because it’s a continuation and acceleration of the 3 changes we discussed earlier.

 

Enhanced by Zemanta

Recent Posts