What does Nutanix actually do?
Nutanix has cracked US$100m in quarterly revenue for the first time. The magic number - $102,697,000 to be precise, emerged in amended pre-float documentation it has lodged with the United States Securities and Exchange Commission. The amended Form S-1 now includes the quarter to January 31st, 2016. That quarter was a beauty …
Basically, instead of having feature rich specialist equipment such as a SAN which are designed for the task and which offer high performance and data protection they allow you to use "off the shelf" server hardware (usually still proprietary vendor stuff like HP or Dell) to create a system which is just about good enough for the task but doesn't need specialist knowledge to configure. This often leads to data consistency problems which are never identified because of this lack of specialist knowledge.
Ironically, this is the same issue as shadow IT - people who don't know better are being sold rainbows and unicorns with a marketing campaign which essentially pretends this tech is cheaper, better and easier than what they have now. Even a cursory examination will show that it's not cheaper or close to cheaper, it's not removing vendor lock in, it's not actually faster if you value your data, and it's not easier in any but the most simple cases.
All that said, the figures would suggest there are plenty of "IT Pros" looking for exactly this technology so good on Nutanix for meeting a "need".
i) Have you ever actually used Nutanix?
ii) What specific data inconsistencies (or other issues) have you encountered?
I'm not an evangelist, but have had to test numerous private cloud platforms (converged and hyper-converged) and found Nutanix rather fabulous, so would like to know of any platform problems.
The same data consistency issues faced by any single controller networked disk storage solution. There are two methods possible, either confirm the write when one controller has the data for low latency but at the risk of data consistency problems, or confirm the write when two controllers have the write which gives worse latency/speed than traditional SAN but does do proper data consistency. There is a lot of stuff in traditional SAN to ensure you don't lose or corrupt data, and SDS addresses very few of these things and worse hides others from the admin. This isn't a Nutanix issue it's pretty much all the converged solutions because they all rely on SDS.
In order to mitigate some of this, you can put in expensive networking and higher powered hosts. You've already bought far more disk than traditional SAN would need due to the network RAID requirement, so overall you end up buying more hardware than you would have with traditional solutions.
Converged works well for VDI - here you don't give a monkeys about consistency of data and you need standard building blocks with known resources. For everything else, normal hardware is still better overall and nobody outside marketing would believe otherwise if they actually a) knew the subject and b) took the time to investigate pricing and c) took the time to discover how the whizbang solution works.
For the record, I have not encountered inconsistencies - because I understand how the solution works, so I'd never put anything important on it without a good reason. Your experience is pretty normal, everything looks fabulous until it isn't. You clearly have yet to hit a problem scenario such as power loss or other sudden host failure and I hope you never do because at this stage I assume it's too late to put something in which will protect your data. In the mean time, buy a really big UPS and configure it properly and that will protect you a little.
How was that wrong? I covered that scenario above and stated quite correctly that this gives worse latency than traditional SAN due to all the communications involved. Traditional SAN has two controllers in the same box with coherent caches, because traditional SAN was designed specifically for the purpose.
"If you check the Nutanix Bible (nutanixbible.com)"
Yup, that would be the same marketing Nutanix uses elsewhere. It suggests that Nutanix removes the network latency of SAN by placing the storage locally which cannot happen with acks to two boxes connected by network. Well done for quoting marketecture in a technical conversation though...
Sorry to disappoint your hate, but Nutanix has data locality for reads which normally does not go over the network. For writes just like SANs as they also must go over the network.
Not all SDS solutions are they same and there are inferior solutions out there that can give you all sorts of data resiliency issues. Nutanix was designed to keep your data integrity intact, resilient, and survive failures.
At least Nutanix takes the tact of providing the information about how the system functions versus complete obscurity through complexity. Why not give the technology a shot versus just placing the blinders on and siding with a company that's going down the drain because they couldn't innovate and understand the shift of the industry?
Tell you what, come on over to our expo, ask the experts, see the technology in action, and then make the decision on whether you'll keep siding with the dinosaurs or embrace the evolution of the datacenter as everyone except Netapp has...