HP's new P10000 3PAR Storage System array, announced Tuesday, is both more and less than expected, with a new ASIC but no SAS drive support. Array federation across 3PAR and LeftHand arrays was also announced. The P10000, or 3PAR V-Class, is a high-end 3PAR array supporting up to 1,920 drives. These are 3.5-inch format Fibre …
I sure do hope these fly off the shelves lest HP chuck a hissy fit and can the whole product line.
Triple digit growth
Well the article yesterday said 200% growth in 3PAR so you could say 3PAR is flying off the shelves!
Good things come to those who wait ( but sometimes very late )
3PAR started the development of the V-Class at the end of 2005, see the link
It is a PCI Express Gen1 design which can be verified with the 2GBps speed of the inter-controller links. A x8 PCIe Gen1 link allows 2.5GBps, but all 3PAR designs have a slightly lower inter-controller link speed. The links are bidirectional which means 4GBps in effect. Therefore we have a backplane bandwidth of 112GBps for the V800 and 24GBps for the V400, the formula is n(n-1)/2*4GBps ( n = number of controllers ).
The V-Class controller has two fourth generation PCIe-based ASICs and six PCI Express Gen1 busses with 2.5GBps, this results in a controller bandwidth of 15GBps for front-end and back-end data traffic. Three quad port FC cards share two PCIe Gen1 x8 busses via PCIe switches which is not so much for the coming 16Gbps FC and 40Gbps iSCSI / FCoE protocols.
Im my opinion it is impossible to use the V-Class technology for a F- and T-Class uplift because of PCI-X / PCIe incompatibility, especially for the backplane. It is more realistic to await a F-Class follower system ( G-Class ??? ) with one fourth generation ASIC and a halved number of PCIe busses per controller . For the more modular G-Class the SAS backend is a must.
You can found the new V-Class architecture guide at
This sums up the problem with hp marketing.
".......a capability that federates arrays to provide what HP describes as "distributed volume management across self-governing homogeneous peer systems using native communication between peers". The idea is to start linking multiple storage systems together and have them act and be managed as a single entity....."
The first sentence is the hp veep's self-important waffle, the second is actually what he should have said.
As regards the 3PARS, still a bit ho-hum to be honest. The XP arrays already do "federation" through having external storage, only a lot better because the external storage can be a whole range of hp and non-hp storage arrays, rather than just one.
What happens with the HP P9500 now ?
The P10000 announcement of HP 3PAR is a big affront against the Hitachi Ltd. , which is the OEM supplier of the P9500 ( Hitachi RAID700, identical with HDS VSP, but provides only small form factor SFF 2.5" disks ). The 9500 plus 500 equals 10000 signals the positioning of the new 3PAR V-Class above the current Hitachi RAID700-Class which is very arrogant or is a "hidden message that the V-Class supports only large form factor LFF 3.5 disks" ( meant ironical ) .
We have seen an interesting article of the HP blogger Calvin Zito aka @HPStorageGuy containing Dave Donatelli's famous "four pyramide picture" presented at the HP Converged Infrastructure Event held last October in Barcelona, look at
for this. The "four pyramide picture" contains the clear message that only a small mainframe & FICON niche remains for the HP XP and P9500 in the future. Therefore the P10000 announcement is no good news for the Hitachi guys, they loose after Oracle Sun the second OEM partner. This should shrink RAID700 sales numbers for the Hitachi Ltd., but can be an advantage for HDS which is becoming more important for the mother company and must compensate the losses. But I think that a net loss remains.
When I see the HDS Roadmap foil from Chris Mellors article "Hitachi goes XYZ over next-gen storage" I think the Hitachi guys have not realized this disadvantageous situation or will only "confuse the Russians". There is no time for going over XYZ and long-term roadmaps should be a secret. The Hitachi Ltd. has prolongated the OEM agreement with HP last year, this was under the current circumstances a big strategic mistake because attack is better than defence.