Converged systems market is so hot it just went backwards Converged systems are supposed to be the hot spot of the otherwise-troubled server and storage markets. Yet sales just dipped for 2016's final quarter, according to kit-counting firm IDC, and overall growth for the year was tepid. The analysts' Worldwide Quarterly Converged Systems Tracker for 2016's final quarter found that …
The whole point of converged systems is to more efficiently utilize your resources. So it makes sense that as they mature and improve the revenue should drop. i.e. if you are better able to use deduplication and compression in your storage you buy less (not to mention how quickly the price of flash is falling)
If it wasn't for greater efficiencies no doubt the storage growth would have been a lot higher than 4%, because we seem to keep finding ways to generate more data. We just don't have to keep 10 copies of it anymore.
We've had dedupe and thin provisioning for ever, it's not a new thing,
Suddenly selling your storage as bundle with a server doesn't make it magically more efficient.
People are "meant" to be moving away from stand alone solutions to converged therefore the market "should" be growing, do you really believe it's not because storage has suddenly become more efficient?
You need to remove your head from the sand and look around, why do you think AWS and Azure are growing so much and the infrastructure vendors aren't?
Less and less care about servers or storage, they care about their applications.
Enterprise disk arrays didn't have anywhere near the controller horsepower required to perform dedupe and compression across all LUNs on a busy array. That didn't happen until all flash arrays made their appearance. It was out of necessity - flash cost so much at first they needed to reduce the amount required to an absolute minimum.
Or maybe you can point me to an enterprise disk array introduced more than five years ago that offered dedupe & compression for all LUNs, since you think we've had it "for ever".
Sure, hyper-converged has some great features. But i've never understood the pricing strategy of companies like Nutanix. They seemed to have missed the opportunity to disrupt and seize the market, rather they priced at a price point that might have stacked up in an TCO comparison against the traditional players, but they forgot they weren't an established player.
hey take $20k worth of hardware, throw their software on it, and flog it for $120k. Sure the software is cool, but I can't help but think that they're have far more impact on the market if they priced the bundle closer to it's actual cost. Just because it's 20% cheaper than the HP / EMC / Lenovo/ Dell option on paper doesn't mean it's a bargain and people will flock to it.
Also, the big players like HPE have enough margin to offer very competitive pricing to large buyers, whereas Nutanix really need those margins to stay in business, meaning that the pricing can be very close to parity. In that case, the buyer is banking on Nutanix offering sufficient added value over a mainstream solution, and that's by no means certain.
The other thing about non-converged solutions is that they're modular. I can easily buy my compute nodes from different vendors and then point them at a variety of backing storage arrays based on my particular needs. With hyperconverged, you're locked into a single vendor's product, one which may not be suitable for all workloads.
The bubble has burst for HCI nonsense. There is only one real use case for this - ROBO with no or very limited local IT skills and a fixed workload. HCI for the enterprise? Don't make me laugh. Who wants vendor lock in to a solution that does not independently scale?
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
