It's May and an time for HP to refresh its evergreen EVA storage line. HP said it would keep investing in the EVA when it bought 3PAR and has kept its promise, with two new models being announced, with larger drive support and better management SW. Being "announced" means HP storage product marketing manager Matthew Morrissey …
Does it still use a processor slower than the one in my BlackBerry?
Lack of mid-range
HP has such a jumble of storage products. I am sure they want to dump EVA but they don't want to lose the install base and they don't really have a replacement mid-range array. They have decent low end with Lefthand and the same LSI stuff that everyone pushes. A good low end of the high end system in 3PAR. XP is good high end system, unfortunately it is Hitachi's high end system. Nothing really in that mid-range space to mess with VNX, HUS, V7000, FAS. EVA is the traditional competitor in that space, but it is a generation behind the competitor's systems.
Stop throwing money down EVA...
and focus on 3PAR development. Errrrbody knows that EVA and XP will eventually be replaced with 3PAR.
HP do not use LSI for the entry level they use Dot Hill. Lefthand has fallen behind Equalogic as far as modular storage goes and it is going to take some time to catch up again (autotiering).
They badly need an entry 3Par product to compeate with the V7000, VNX and HUS.
XP is the only HP option if you need FICON, however P10000 gets a LOT more focus, and makes more sense, from HP unless there is a FICON requirement.
I have heard rumors of a new MSA branded box from Dot Hill that supports auto tiering and useing the P2000 G3 as expansion in the mid-year timeframe.
XP is going to be X-tinct
XP is going out. As you mention, HP is trying to migrate XP users to high end 3PAR and not mentioning XP to any prospectives. They are not going to continue OEMing Hitachi making OEM margins when they can sell their own gear. The FICON requirement is HP's reasoning for keeping XP, but that is not a major issue for HP. Mainframe shops use either high end IBM DS or EMC Sym 90% of the time. HP probably only has a handful for mainframe shops with XP as the back end.
Re: XP is going to be X-tinct
"....HP is trying to migrate XP users to high end 3PAR...." Agreed, we've been the subject of the 3PAR schpiel. It makes sense that hp would want to switch those customers they can from XP to 3PAR as they make more margin on their own 3PAR. But.....
"....The FICON requirement is HP's reasoning for keeping XP...." It's a reason, but another you forgot is that XP offers a very simple way to consolidate multi-vendor SANs through the External Storage capability. It's also a neat way to go to a simple (but pricey) DR solution if you have mult-vendor SANs. IIRC, 3PAR does not seem to have an external storage capability to match the XP's.
One thng EVA does give is a large number of discs cheaply off one pair of controllers, which 3PAR doesn't really do, and the MSA can't quite scale to yet. I hear hp are relaxing the configuration rules for 3PAR to get round some of the issues of having to have so many shelves per controller pair because it was making 3PAR configs pricey compared to the old EVA. But at the end of the day the EVA is so popular because it's relatively cheap and just so easy to manage.
Re: XP is going to be X-tinct
I am not familiar with the external disk virtualization capabilities of XP, but that it seems that buying an XP would be an expensive way to go about virtualizating third-party disk if that is the only reason you are buying it. Just buy an SVC head or NetApp's V gateway.
End of the day, HP isn't going to hold on to XP. They have to sell 2-3 of those systems to make the same about of profit they make selling one equivalently sized 3PAR system. Even if they lose a large group of their XP users in the migration to 3PAR, it still might be more profitable to migrate them all to 3PAR even with a smaller install base because they are making owner margins instead of resell margins. They are trying to get as many voluntary migrations as they can before they make it an involuntary migration. Hitachi must be pissed off about this situation. Oracle just booted them from the Sun high end (apparently Oracle thought having no high end offering is better than a Hitachi high end offering). They have to be out in those XP accounts trying to get USP-V migrations.
How many customers with USP-V / XP24000 or VSP/P9500 actually virtualise external storage for extended periods (e.g. longer than migrration)?
The reason I ask is IBM's view of this is less than 10% of USP-V customers actually use the virtualisation capabilities of the box and I would like some vendor FUD free info if any exists.
Re: Michael Duke
We already had some XPs attached to UNIX and mainframe before we swallowed up a competitor two years ago. Their SANs had grown explosively, with a new array from a range of "approved" vendors for each project, and they had just thrown money at projects to keep up with customer demand. Their purchasing department used to be able to switch deals to whichever vendor was giving the biggest discount of the day, so they had a real management headache along with a support nightmare. When we bought them I had to go work out what was on their SAN devices and what was actually in a viable DR setup (we have a rule that everything has to have a DR failover/backup to a remote site). The answer was very little.
Given the massive amounts of different devices and the mix of data, consolidating it in the short term was going to be a massive headache. So we bought two XP24000s to act as centralised SAN devices, one per site, setup DR replication between them, then attached the other devices to them as external storage. Not cheap, but it did the job neatly and with the least risk. We then migrated data off the older devices or the ones we just didn't want to keep, and kept a load of others seeing as they were simply cheap and had time left on their support contracts. We still have the setup in-place, we have recycled some of the external devices into third tier archives, and we're scaring the individual vendors into nice support discounts by talking about migrating the lot onto 3PARs - unlikley to happen, but they don't know that! ;)
Let's be honest, many customers who have XP's didn't need them. HP discounted XP's down because the EVA didn't scale. And the market for FICON attach is tiny so the genuine market for the XP is also tiny. 3PAR will offer almost the same levels of reliability (comfortably in excess of 5 x 9's) but with a whole lot more performance, flexibility and ease of management and at a lower cost! I can see HP selling less and less XP's. HDS is on its' arse! And then Hitachi will pull investment. Monolithic storage is a dying beast!
FICON is tiny as compared to distributed, but...
it is also the highest value, highest margin area of storage. Those are the "seriously, if this workload goes down, there is going to panic in the streets" workloads. Having written that, FICON connections have little to do with XP as the vast majority of the FICON world uses either DS8 or EMC Sym. XP can't have more than a handful of IBM mainframe shops that, for some unknown reason, decide to run Hitachi storage with a HP label. The XP shops are usually HP server users... without the vendor consolidation benefit, people would just buy Hitachi if they wanted Hitachi.
All technical arguments aside, HP will drop XP with a purely financial justification. They make much higher profit margins selling 3PAR, even though it may cost them a handful of high end customers, than continuing on with XP. They can lose a substantial number of customers and still make more in profit but just pursuing 3PAR.... Having written that, trading market share for higher margins has always been a difficult concept for HP to grasp... so who knows, but it would be the rational decision.
Re: FICON is tiny as compared to distributed, but...
Wunderburp, you're looking at the whole thing the wrong way round as usual.
".....XP can't have more than a handful of IBM mainframe shops that, for some unknown reason.... The XP shops are usually HP server users...." The reason is simple - there are a LOT more non-mainframe shops than mainframe shops, and ALL those mainframe shops also run critical apps on other servers (UNIX, x64). And seeing as the majority of those non-mainframe customers are going to be using hp servers, it means hp has a bigger chance of being in on the storage discussions. It makes sense to use a storage device that can work with both mainframe and the other servers, and XP (in my experience) simply does that better.