What an amazing three days it’s been at HP Enterprise (HPE) Discover in London! The event was an incredible experience and I am grateful to HPE for my ticket and to my employer for allowing me the time to come to the event. I should probably thank my parents for letting me crash at their place to save having to pay for a hotel.
I have so many things I want to write about and I’m sure I will forget some of them but to start with this post will cover everything that springs to mind right now.
Synergy is a new HPE product which sounds very interesting. It is based around the concept of composable infrastructure – further details here http://www8.hp.com/us/en/hp-news/press-release.html?id=2126684#.VmCgTXbhDIU
I strongly recommend you check Synergy out – while you can’t buy it right now and the initial release will be (as they say) a ‘1.0’ version it certainly has potential and I can see it empowering organisations and users to achieve great things.
MoonShot is a technology I followed when it was first announced then drifted away from as I didn’t at the time have a use case for it that wasn’t already met. Having got my hands on the kit and talking to the engineer for quite a long time I think there is a potential for leveraging it in a virtual desktop (VDI) setup. It all comes down to what the most cost effective option is for us as a business and whether MoonShot provides a compelling enough argument.
The backup solution used in my current role is (I believe) coming up to the end of it’s 5 year life. Obviously for us this means tendering or extending the life of the current product. I really liked what I read and discussed with respect to HPE StoreOnce appliances and the general ecosystem that exists around them. I can certainly see a use case especially considering the integration it has with the 3PAR array. I think it would be great to do a proof of concept to see whether it is the fit we are looking for and a potential option for the future.
I was in absolute heaven at Discover – they had more 3PAR arrays than I have ever seen in one place. Unfortunately I couldn’t quite run off with a 20850 all flash array without somebody noticing. Next time (if I get to go to another Discover) I’ll remember to bring a bigger backpack!
Anyway back to 3PAR – I’ve known about the new replication mode called asynchronous streaming and from what I read and watched I think it will really help in our setup. I knew it would likely be exactly what we were looking to deploy for certain use cases but after this week it’s a definite. The only thing we need to do is add some more ports in a remote copy setup to allow us to do other replication modes alongside synchronous.
Something else I hope to see in the future with 3PAR is NVMe (non-volatile memory express) support for flash storage. This would certainly drive flash performance to new levels especially if they moved from 12Gb SAS.
With regard to storage on 3PAR it looks like the capacity in flash drives is set to continue to grow rapidly. This is great and with HPE driving down the cost of flash the utopia that is an all flash array comes closer for all.
Gen5 Fibre Channel
Fifth generation FC running at 16Gb is a wonderful storage fabric and the reduction they have realised in latencies (end to end using fifth gen) along with the doubling in bandwidth certainly helps to fully leverage flash storage. It really is important to remember that buying a 3PAR 8450 full of flash drives and then connecting to it over a poor fabric will not result in the best results. The choice you make in fabric will have a noticeable impact so make sure you don’t limit your storage array. Personally I would love to see us move away from FCoE (Fibre Channel over Ethernet) to true native FC end to end from the blades to the 3PAR. This would require a great deal of investment on our part in terms of new hardware and then the administrative effort for myself and colleagues to reconfigure. That being said I certainly think it’s worth putting forward to management as our deployment of flash increases.
Server Side Acceleration Cards (Flash)
For a long time now there have been PCIe flash cards designed to accelerate applications on the server side. Our current VDI deployment relies on the 3PAR to provide storage and while this works great we could leverage a FusionIO card (or some other brand) in a server to provide local high speed, low latency storage. As we use ‘pooled’ VDI sessions we don’t have to worry about protecting them so a single local PCIe flash card would not create a single point of failure (SPOF) risk.
I intend to discuss this further with the team as we plan our future VDI build for this business.
It’s also something I would like to suggest be evaluated for some of our database environments to accelerate their workload. One that springs to mind is our business intelligence (BI) system. This takes many hours to complete it’s work and I wonder what value the business would place on a reduction in processing time?
There are actually quite a few other topics I could talk in depth on around security products and other fabric/compute/storage enhancements but I think for now this will suffice.
If you went to Discover what were your takeaways from it – what are you excited about? Drop a comment or message me.