Will You Put the Data in PernixData?June 25, 2015
CTO Satyam Vaghani was kind enough to announce several new products and features relating to the future of PernixData at Virtualization Field Day 5. If you’re not familiar with PernixData, they got their start with their FVP product which provided server side flash for both a read cache or a write-through cache. I’ve used the product several times and it really does some amazing things to smooth out latency and can give your storage array some serious umph!
PernixData presented on June 24th at VFD5 where I was lucky enough to be a delegate. Satyam announced four new updates.
All travel expenses and incidentals were paid for by Gestalt IT to attend Virtualization Field Day 5. This was the only compensation given and did not influence the content of this article.
PernixData Architect is a new product that provides insights and deep analytics about workloads running in your infrastructure. The sales pitch here is that its used to help design new infrastructure implementations, and then monitor and optimize the design after its deployed. Utilizing deep level statistics will better allow customers to decide how much flash should be bought, where to put it etc. Once deployed, what-if scenarios can be run to help optimize the existing infrastructure and determine how to make things run better. PernixData Architect is doing this by adding an agent to each ESXi host and collecting its own data. Satyam was adamant that any statistical metrics gathered should be gathered by PernixData’s solution and not a third party system. He went on to explain that its dangerous to use some other system’s information repository (such as vCenter statistics) because they either might not have the statistics necessary to make recommendations or collect data slightly differently causing a butterfly affect of misinformation. Its refreshing to see someone take this approach as I’ve seen issues trying to marry array statistics and vCenter statistics together.
This is also a new product but works in conjunction with the PernixData Architect product. The Architect software is collecting data about your environment in order to make recommendations on design and optimization. PernixData Cloud allows you to anonymously upload this data to a public repository to then compare against all of the other installations of the product. This would allow you to compare your environment with a global pool, or some categories such as vertical, or company size etc.
I can see why it might be nice to see what the consolidation ratios are of other customers in your space, but are you really getting the information you’re looking for? It seems to me that unless the product really takes off, you’ll only be comparing your infrastructure with the infrastructure of other customers who would need PernixData FVP. I would suspect that any customers who have an aging storage array and need additional IOPS or lower latency can buy PernixData FVP to fix the problem. This is great, but if you look at the other customers data, will all of them have aging storage arrays that need additional IOPS or lower latency? This means that in order for the PernixData Cloud solution to be a real value, customers will want to buy and install this even if they don’t need FVP.
I brought up the question about not having enough data to make an accurate comparison with an industry. Think about it, “Big Data” isn’t really big until you get a lot of it. A little data isn’t a very good representation of the collective industry. To combat this problem PernixData is announcing “Freedom” which is a new program that allows you to use the FVP product for free. The caveats are that you can’t use the local flash devices and are forced to use the Distributed Fault Tolerant Memory (DFTM) version. This allows you to carve out ESXi host memory as a flash cache. This is also limited to 128GB per cluster for write-through acceleration, but thats still really cool.
OK, there is one last catch with “Freedom”. If you use “Freedom” you also have to provide your anonymized analytics information to the global repository to be shared with other customers. This is a pretty small price to pay for some really useful software, but it might be a sticking point for some companies.
Does this “Freedom” solution fix the issue of getting a large amount of “Big Data” for analytical purposes? You could argue that the customers that are going to use this free software aren’t enterprise customers anyway. Will this software only be installed in home labs or development environments? If so, then comparing your own infrastructure with others isn’t really getting you what you want. You’d be comparing your home labs against each other right? Will “Freedom” determine that a high percentage of storage arrays are manufactured by Synology? Only time will tell.
Last but not least, there are going to be updates to the existing version of FVP including things like vSphere 6 support, a new User Interface and pulling it out of the vSphere web client.
I applaud PernixData for trying to branch out and do something more with some of the data they’ve already been great at collecting. I think this could be a useful product and they really do seem to show some VERY useful data about storage fingerprints. I’m skeptical about whether or not the PernixData Cloud / Freedom messaging will work out, especially since its only focusing on storage for now. If the industry takes to the software and adopts it with several large enterprise customers I think it’s got some very useful information that IT departments would love to see, but if not then I don’t know that we get the comparisons that we’re hoping to get.
If any company can make this work, then PernixData can. The initial analytics show great storage fingerprint information and the FVP product is a rock solid solution. Lets hope the new products do as well as their flagship product.