Original URL: http://www.theregister.co.uk/2009/11/03/ecv_v_block_vs_hp_cell/

EMC/Cisco's V-Block faces the hard Cell from HP

Integrated stacks to face off

By Chris Mellor

Posted in Storage, 3rd November 2009 11:21 GMT

Comment EMC and Cisco have announced a plan to sell virtual blocks - or V-Blocks - likely to be integrated stacks of virtualised servers, storage and switches, either as products or services. HP's Cell technology could achieve the same end: IT stacks provisioned on demand in private or public clouds.

A V-Block is, El Reg reckons, a virtual block of IT gear. It is a basic set of server, storage and switch resources described as a metadata set and capable of being provisioned, operated, managed and de-provisioned by a data centre operating system, a vSphere. Cisco and EMC bring the server, storage and network building blocks, and VMware brings the virtual glue to bind them together into a usable virtual machine inside a virtualised data centre.

We reckon V-Blocks will be able to communicate with each other and interoperate, with policies setting processing needs, network type and bandwidth, and storage quality of service, which would detail capacity, media type, protection method and so forth.

Nobody else, no EVC (EMC, VMware, Cisco) competitor, has this, or do they? The present competitors are Dell, HP, IBM and Sunacle (Sun + Oracle). Players that could join this integrated and virtualised data centre IT stack party are Hitachi and Fujitsu. They have the physical pieces but not the virtualisation software. For that they must be looking at Citrix, RedHat and Microsoft, with the latter being the main hope.

Dell is not, let's say, a true enterprise-class competitor to the others. It's smaller than EVC, HP and IBM, doesn't have its own big iron, server or storage, and it is still building up a top-level services capability. Sunacle is also limited in scope as it's going to be focussed on Oracle applications and is currently going through a Sun transplant surgery and recovery process that could take years.

That leaves HP and IBM. IBM lacks an in-house networking box capability and, so far, has shown public reluctance to building its own integrated IT stack offering and naming it, like Cisco's UCS California concept, or HP's Blade Matrix. Of all these EVC competitors, HP is the closest with its own branded integrated data centre stack offering, BladeSystem Matrix (BSM), built from in-house pieces, although using VMware.

Derek Cockerton, HP's director for BladeSystem, Software and HPC in EMEA, says BladeSystem Matrix is: "a shared services infrastructure delivered to your data centre... a cloud in a box." You can "provision from shared pools with a self-service portal... You need real-time charging to make cloud real. BSM will get real-time billing soon."

HP has: "taken six orders for BladeSystem Matrix in Europe," by the way.

BladeSystem Matrix is at the heart of HP's virtual data centre (VDC) concept. The brain though, is the management layer, software that's broad enough to cover the entire data centre and deep enough to go into the individual organs inside the infrastructure, and treat everything there as just another VDC component.

Virtual Data Centre cells

The VDC is composed of a pool of resources, servers, storage and networking boxes, with all three bladed - HP's VDC mantra is to blade everything - treated as virtualised resources and capable of being deployed on a massive scale, and managed.

Management is key. VMware has its vSphere, which most likely will become this for ECV, but HP has Cell ideas in its HP Labs.

Cell technology treats VMware as just another VDC component, and wraps it up inside a quasi-VDC operating system based on sets of virtualised machine resources called cells.

According to John Manley, Director of Automated Infrastructure in HP Labs, Bristol, a cell is an infrastructure-level cloud service, a secure and isolated partition in the infrastructure - a virtualised entity. It is, as you would expect, composed of servers (HP blades), switches (ProCurve) and storage (XP, EVA, etc.) operated as virtualised resources and described by a set of metadata, a model.

The model data describes - or rather specifies - the cell's loading, response time needs, scaling behaviour, failure behaviour, security, connectivity and so forth.

Models are cell templates and describe every necessary component, applications and operating systems included, for a cell to be loaded and started up. You could have, for example, a database cell, an order processing cell, or an accounts payable cell. The cells are virtual black boxes.

Once running, cells can be combined to form business processes. Cells in a data centre can be extended to other data centres, or to service providers to do this. They are virtualised building blocks, V-Blocks in ECV's language, for horizontal and vertical cloud services.

The management layer hides the complexity involved and provides life cycle management of cells. You would use this to build a customised set of services in a virtualised data centre, for IT service consumers in public or private clouds, or both. Manley reckons that the cloud will be absolutely heterogeneous in future, and will use federated domains. (Find out more here (pdf).)

Service providers could use cells to deliver services to multiple customers simultaneously. We could think of cloud service providers as operating cell factories which run on large scale, modular, flexible and virtualised physical infrastructures.

Controlling HP's destiny

With this cell-based management technology in its labs, HP has been developing cloud infrastructure management layer software that could look after a variety of virtualised server environments, VMware, Hyper-V, Citrix or RedHat. By developing its own virtual data centre management product, HP would avoid being in thrall to VMware or dependent on Microsoft's schedule for developing and extending Hyper-V.

It would be far better for HP to be in control of its own destiny and this HP Labs Cell technology indicates how it could do so. Were HP to make this into a product, then it's arguable that it would be making the running in the virtualised cloud data centre stakes and not EMC and Cisco, who would be relegated to catch-up mode.

In fact, it seems that we could say, in terms of virtual data centre deliverables, HP is actually ahead of Cisco and EMC already. Neither of them, on their own, is capable of challenging HP in this area. They have to partner, having no other choice if they want to compete, and EVC and its V-Block is the result.

These two individually weaker players, in the cloud VDC stakes, are partnering to challenge HP, and then IBM. That's one way to look at it. We'll just have to wait to see whose offering comes out on top. ®