Hi and welcome to EMC World 2013. For those that are unaware what EMC World is: It’s an annual seminar covering all things new and improved that EMC has to offer. It’s hosted in The Venetian hotel in nothing less than … Las Vegas! Your ideal location to learn about the new products EMC is launching, to socialize with fellow EMC customers, partners or employees and of course… to have some fun! 😉
Yesterday we checked in to EMC World, got our access badges and goody bag. This being the first actual day of EMC World, we kicked it off by immediately leaving the premises!
As you may be aware, EMC formed a group of social media evangelists that share their technical expertise and talk about EMC products and services. This group is called the EMC Elect. Each year people can nominate people to be considered for the EMC Elect title. For the batch of EMC Elect 2013 over 200 people were nominated of which 75 people made the final selection. Two colleagues (Menno de Liège and Rob Koper) are included in this elite group of social media addic.. gurus, with Rob even being one of the original founders.
This morning’s roadtrip was one of the perks of being EMC Elect: we went off to see the Switch datacenter SuperNAP. SuperNAP is one of the largest and definitely most secure datacenter in the US, located in Vegas itself. For us Europeans a nice chance to see how a US / Vegas datacenter operates.
Our host for the day was
@MissyByte, which told us about the mission of Switch: provide 100% of uptime, all the time. You can see this in every design aspect of the datacenter, from power to cooling to security and fire safety. In the picture above you can see the N+2 power redundancy: everything is color-coded so that you can make no mistakes when plugging in equipment. Next to that the entire datacenter is one cold area, with hot air sucked out of the back of the servers and up into the ceiling. Custom designed chillers that choose the most efficient way of cooling depending on the outside weather… the list is endless. Check out their website is you want to know more!
Back at EMC World, my first breakout session was about VPLEX, specifically how it functions in conjunction with VMware High Availability and VMware Fault-Tolerance. EMC VPLEX is a storage virtualization appliance. On one end you attach all your storage systems, on the other end your servers. VPLEX virtualizes your storage, which has several advantages. For example, if you wish to phase out an old, end-of-life storage system, you can do so without disruption to the attached server. VPLEX will migrate the data to a new storage system, without the server ever noticing.
On the other hand (and that was today’s subject of the breakout) it offers increased availability. Let’s say you’ve attached a LUN on storage A to a server. This LUN is replicated to a second storage B. Should storage A fail you can restart operations on the secondary copy on storage B. The vital word here is restart. You will always have some kind of downtime, e.g. an RTO (Recovery Time Objective) that is larger than 0. With VPLEX this is no longer the case. VPLEX will handle your replication (it will split the writes once it receives them from the server and sends them to the storage systems you have assigned to host the data). If your primary storage system fails, VPLEX will switch-over seamlessly to the secondary storage system.
Combine this functionality with the features VMware HA and VMware FT have to offer. VPLEX makes sure that a storage failure doesn’t impact your operations. VMware FT does the same for the computing layer: lose an ESX server and the mirrored virtual machine will continue to run on a different ESX host. Combine the two (VPLEX) and you’ve got datacenter redundancy: no downtime, no data loss, even when your datacenter goes down. Pretty awesome, right?!
One word of advice: upgrade to VMware 5.1! There are a large number of improvements made to limit the impact of APD (All Paths Down) and PDL (Persistent Device Loss) events. Pre 5.1 a loss of all paths to the storage would cause your ESXi host to become unresponsive. VMware HA would not trigger and you’d have to manually reset the ESXi host.