P10000: What HP wants to make instead of PCs, tablets Recent rumours have renewed storage industry interest in a forthcoming topping-out of the HP 3PAR range with the P10000. We first heard about a new, unnamed box in July, but sources are now telling us about a P10000 which will slot in above the current range-topper, the 800TB raw capacity T-Class and its smaller brother, the F …
They key word here is "Sourced"
It is a natural progression of things, when excel mongers drive a company off the road (as per http://www.time.com/time/magazine/article/0,9171,2081930,00.html) and replace "high cost" functions like engineering, design, etc by sourcing and clever pricing manipulation. The company usually undergoes an initial "jump" in its financials. Money is no longer wasted on "smelly engineers", innovation is outsourced to OEMs, the company is "lean" shop-n-ship shop and we can all sing, dance and enjoy the stock price rise. That lasts 5, 10 sometimes 15 years.
That is natural as this is the time it takes for the market to mature and enter into margin squeeze. After that period the margin of the shop-n-ship is squeezed and the shop-n-ship portion reduced to a "race to the bottom" scenario. At that point the only place where you can generate margin is by designing something new and by creating new product.
However, if you have "leaned-out" and sourced all of your product you no longer _OWN_ that margin. It is with the OEM+ODMs. So you have to live on shop-n-ship margin which is guaranteed to be reduced long term in a competitive market.
That is the trouble with HP - the margin-carrying parts of its PCs are now largely designed by Asus (and few other ODM/OEMs), its storage is designed by HDS, its... So it does not retain any margin from new products and its margins are only from lean shop-n-shipping which is too cut-throat.
That is why it financial results suck in an area where Lenovo, Apple, etc have managed to excel. They have kept their designs. No "sourcing". So in a market where the "shop-n-ship" part of the margin has dropped to under "commodity" levels they can still make money and lots of it.
It is a classic example of a rise in market share followed by a fall through the elimination of "what gives your company a future".
the raw capacity limits in the current 3PAR systems is all about the software - which is 32bit and limits them to 4GB of control cache per controller. This new box is supposed to have 64-bit software which will remove that limit, though I've been told in one way or another 64-bit has been coming since 2008(having needed to get past that limit at that point in time), maybe this time it'll actually happen.
Then there is the federation stuff that was first mentioned on the register about a year ago.
I'd be expecting overall I/O performance to more than double vs their current gear, and I get hints that they aren't going to drop the current T class systems and replace them with the newer boxes, but rather have the newer dual ASIC system be a totally different tier of storage.
I hope they back port the 64-bit code to their existing systems but I fear they will not (not because they can't since existing T/F systems have 64-bit processors but because they won't - if 3PAR was still independent I'm sure they would back port).
Both the T and the F have held their ground amazingly well in my opinion given how long they've been on the market, they're still the fastest single system disk arrays out there.
In general I am surprised more storage companies have not adopted the distributed sub-disk RAID technology available in some form or another in systems like 3PAR, XIV, Compellent and Xiotech. Clearly it improves performance and lowers latency as well as reduces RAID rebuild times and reduces risk of data loss because of multiple disk failures. XIV's hardware architecture sucks, wonder why IBM hasn't fixed it to make it more scalable.
I was looking at the I/O performance of a NetApp system recently and was just in shock as far as how poor it was, and the lack of ability to run anything other than RAID DP. Me - I'm used to running RAID 5 3+1 on 3PAR which is only 5-7% slower than RAID 10, and at least 2.5x faster than NetApp on RAID DP with the same number/type of spindles. I've been using 3PAR so long I've just come to expect that level of performance, I guess I should try to lower my expectations for when I come across other platforms. The large bulk of the I/O advantages on 3PAR are because of the distributed sub-disk RAID. Need some volumes RAID 10, some volumes RAID 5 on the same disks? no problem.. -- other platforms - big freakin' problem because they can't do it.
HP really has jacked up the prices of 3PAR kit since they bought them and have been throwing in lots of additional support fees that simply didn't exist before, which is unfortunate. But apparently hasn't impacted sales much.
I've been waiting for the G-class systems for a while now just for the marketing opportunities alone...
It might have a 'stunningly simple management interface' but why won't XIV just give you the size of LUN that you ask for? And have they fixed that double disk failure hang problem yet?? Oh, and how many weeks to do zero reclaim on even a small LUN? 3, wasn't it? There are bigger problems it needs to address rather than throwing some marketing performance numbers at us in a youtube video!
The T will run the same code that the P10000 (formely known as the V series) will.
So it will support the same features and will be fully 64-bit.
3PAR has a common hardware architecture so all systems run exactly the same software.
As long as the hardware supports it, all old system inherit new features and abilities.
So HP's CEO wants to sell the PC side of the business, even though its the biggest PC maker in the world. So what kind of a bonus does the CEO get from such a huge sale?
The directors (more interested in share prices and company acquisitions) have built the company (image) up with company acquisitions and now its sell out time, to break up and sell off parts of the business. So big bonuses both ways for the directors, during both acquisitions and sell off's.
So all the talk of the end of PC's is just a smoke screen lie. Its an excuse to try to justify (to shareholders) that the act of selling up the PC making side of the company has to be done. It doesn't have to be done, but then the directors know they will get huge bonuses when it is sold, so they are prepared to say whatever it takes to get that big fat bonus ... then once they have bleed out any remaining value from HP, they can leave to go to another company, to do the same thing again.
Its a reoccurring pattern often found in directors of companies. Its like the old saying, "follow the money". The directors don't really care about the company and its workers, they only care about their big bonuses. The directors are being very narcissistic, but then that's company directors. :(
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
