Who won all-flash sales sash, sucked up all the cash? – IDC report An EMC-distributed excerpt of an IDC all-flash array shipments and revenues report has Hopkinton topping the sales rankings, with NetApp at number four and IBM topping the capacity-shipped chart. The report is a special IDC one, titled Worldwide All-Flash Array and Hybrid Flash Array 2014–2018 Forecast and 1H14 Vendor Shares …
Measuring vendors on a $/RAW GB basis is a useless and misleading exercise to end users. What matters MOST is what the effective usable capacity is to the end user, i.e. how much data can I actually write to the array/appliance, after all hardware, software, and maintenance are included. $/Usable GB at the end of the day is the only thing that matters with the added influence of density, power/cooling, simplicity, and another big one being resiliency and NDU. Pull two drives from all of these vendor's products and see who survives. See who can do NDU code, capacity, and performance upgrades with no downtime and/or loss of performance.
I am not sure why anyone is surprised to see EMC leading any chart that has to do with storage revenues. There systems are ALWAYS more expensive then similarly configured systems from competitors (HDS might be the exception, and with good reason), and they are the least efficient systems out there, meaning you will need more physical storage space than competitors since they require the most disk/flash for whatever you are going to load up on their systems, so of course they ship more raw disk than anyone else. This is on top of them being the largest storage vendor in the world for other dubious reasons.
That said - who really cares about AFA's? What percentage of companies out there really uses or needs them in the first place?
NetApp recently started publishing storage limits to FAS boxes with SSDs in them since some of their customers configured them and bought them as AFAs anyway. This was probably also done to shut up anyone whining about them not having an all flash array.
It is a small portion of their customer base, because few companies require AFAs for their business, and those that do don't need it for everything.
EMCs flash caching solutions are notoriously bad and difficult to configure, and they are all about shaking more money out of customers, so it makes sense for them to push their current customers into AFAs for specific purposes (reporting databases, highly transaction stuff, etc). Not that these wouldn't perform well on a hybrid system from another vendor, its just that EMC doesn't have a good hybrid system and they are already in the position of having locked in customers that buy whatever they tell them to buy, so its an easy sale.
Measuring vendors on a $/RAW GB basis is a useless and misleading exercise to end users. What matters MOST is what the effective usable capacity is to the end user, i.e. how much data can I actually write to the array/appliance. $/Usable GB after hardware, software, and support are all factored in at the end of the day is the only thing that matters with the added influence of density, power/cooling, simplicity, and another big one being resiliency and NDU. Pull two drives from all of these vendor's products and see who survives. See who can do NDU code, capacity, and performance upgrades with no downtime and/or loss of performance.
Hi Liger,
First of all, yes I am biased, I do work in the HP Storage division.. but i'll try to be objective here..
You are wrong.. eg. In a 3Par Storeserv system deduplication, thin persistence, etc are NOT workload dependant as those are covered through hardware rather than an added software load on the controller CPU's. IBM for instance in their StorWize series now does data compression in hardware as opposed to software just in order to reduce the influence on overall performance. NetApp has been adding hardware technologies to improve the performance of their FAS arrays for years (eg. NVRAM)
In my opinion these are the major factors customers should consider when buying arrays :
- Net capacity... Why? Do you really want to pay for stuff you can't use ? Eg. Vault drives in an EMC VNX environment? All the NetApp overhead ? The overhead that is existing in a HP 3par or IBM StorWize? The overhead is different for every vendor and should be taken into account in any storage deal..as it inflluences your TCO a lot ...and it is the only way to get a correct normalization.
- Performance : Most customers don't care if you are offering an AFA or a hybrid array..as long as they get the performance they're asking for..
- Puchase Cost : what do I pay for all of this? This is where AFA most of the times fails.. You still need to be aware that AFA is still more expensive..and a correctly sized hybrid array can give you exactly what you need at a much better price point.
-Cost of managing and maintaining the array..very often overlooked..but in the lifecycle of your arrays this is maybe your biggest cost.
This all being said, I guess the brass at HP storage are now also knocking at the doors of IDC.... to find our where all of those numbers come from.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
