if you get the chance to be at EMC World you probably saw an interesting demo shared by Cisco, EMC and VCE about Mobility and Business Continuance – If you didn’t , Cisco Live San Diego will be another opportunity to see it
Our favorite bloggers Jake Howering and Omar Sultan wrote in the recent past about DCI (Data Center Interconnect) , OTV (Overlay Transport Virtualization) i.e DCI as an enabling framework for both Workload Mobility & Disaster Recovery
Today I am pleased to have EMC Colin Durocher, bringing his perspective on the best way to address a critical challenge for a lot of IT organizations.
Next week I will post a second part (here) , with a video about the demo itself
Colin Durocher (on Twitter @OtherColin) is a Principal Product Manager with the RecoverPoint VPLEX Business Unit.He has been working with the VPLEX product in several capacities including QA, software development,
systems engineering, and product management for over 10 years.
He is a father of two, a professional engineer, and is currently pursuing an MBA.
Colin is based out of Montreal, Canada.
“Life Inside the Datacenter Silo
The traditional approach to IT is characterized by datacenter silos. Within each silo, we have our operations down to a science:
- We use server clustering, redundant network fabrics, and RAID storage to protect against unplanned local failures.
- We maintain spare capacity to absorb failures and workload spikes
- We don’t think twice about moving data between tiers, or even between arrays to optimize cost and performance.
- We commonly move virtual machines non-disruptively from server to server to load balance or perform maintenance.
As far as mobility and availability needs are concerned, life is good… Within the silo.
Crossing the Chasm (Between Silos)
When it comes to protecting against site failures, we use array replication to maintain a copy of all our data in a secondary (often passive) datacenter. We maintain scripts to automate our failover in case we ever need to declare a disaster. We practice our DR plan at least once a year. Don’t we? Moving applications between datacenters is complicated enough that we really just try not to do it. When we do, it often entails a professional services engagement.
All this has worked reasonably well for us up to now. But IT budgets are being squeezed and IT administrators need to eliminate waste, reduce complexity and find ways to increase their operational efficiency. It isn’t an optional thing. Consider the IDC digital universe study (2011) which estimates that by 2020, the amount of information under management will increase by a factor of 50 while the number of IT staff managing it will increase by only 1.5
That gap will need to be filled by different technologies. Let me introduce one to you – EMC VPLEX Metro. For hundreds of customers, it is breaking down the barriers between datacenters bringing new levels of efficiency, simplicity, and availability.
What VPEX Metro does
VPLEX makes the same information available in two places at the same time. It sits between your servers and (heterogeneous) storage in both locations. It virtualizes the storage with enterprise-class hardware and software and creates what is called a distributed volume. VPLEX’s AccessAnywhere technology – a globally coherent distributed cache architecture borrowed from the world of super-computing – ensures fast coherent access to data in both locations. Reads are always serviced locally – from cache or from disk – while writes are synchronously mirrored between locations.
How VPLEx and OTV change the game
VPLEX breaks down the barriers between your datacenters as far as storage is concerned. Cisco Overlay Transport Virtualization (OTV) does the same as far as the network is concerned – stretching a layer 2 network between sites. By combining these, we can now stretch our server clustering technologies between datacenters and stretch their mobility and availability benefits to eliminate the datacenter silos.
Consider VMware vSphere as an example. With VPLEX and OTV, we can now stretch our ESXi clusters across datacenter boundaries allowing us to non-disruptively move applications between our datacenters, just as we do within the datacenter. Given that 80% of downtime is planned, this can be really useful for moving applications out of harm’s way ahead of any maintenance activity… Eliminating the need to negotiate maintenance windows with users.
Even more amazing is that we can leverage VMware HA to protect our applications from site failure without having to maintain any DR scripts or have a human involved to declare a disaster before systems will begin to failover to the secondary site. Also, when the primary site comes back online, we no longer need to take an outage for “failback”. We simply vMotion applications back to where they should be. In fact, with VMware Distributed Resource Scheduler, VMware will do that for us. Self-healing datacenter infrastructure.
Beyond that, we can fully leverage a second site as an active one… In fact, there is no longer the concept of primary and secondary sites. There is now one federated pool of resources. By enabling VMware Distributed Resource Scheduler (DRS), we can even benefit from automatic load balancing of virtual machines across that entire pool allowing us to leverage the spare capacity in one datacenter to absorb failures or workload spikes in another datacenter. VPLEX thus enables better asset utilization.
These benefits are not limited to VMware clusters. The mobility and availability use cases apply to Oracle RAC, IBM PowerHA, Microsoft Hyper-V and Cluster Services, as well as other clustering technologies.
Where to Learn More
Also more Cisco Data Center blogs on OTV