IBM And VMware Find Common Ground In Storage And Server Hypervising

IBM and VMware Highlights

• IBM is VMware’s biggest customer
• Enhances IBM’s role as a storage hypervisor vendor
• Matching approaches to server and storage hypervisors
• 20k common customers
• IBM addresses multiple APIs to make its storage hypervisor relevant within the VMware ecosystem
• IBM Stretched Cluster extends vMotion usage to 300km
• Work together on addressing the cost and latency issues of VDI
• A potentially strong common go-to-market approach

We thought you would find it interesting learn about IBM and VMware’s partnership and our view of how these fit into the developments of virtualised computing; but, be warned, this article contains a mass of acronyms, which we’ve addressed as footnotes. You might also like to see our Acronym Buster for more help.

A Long-Standing – But Less-Known – Partnership

IBM is the largest vendor of virtualised servers and VMware, of x86 server hypervisors. They have around 20k common customers (including 70% of its XIV[1] storage systems customers), while IBM itself is VMware’s largest customer and integrator. IBM currently sells around 250 VMware SKUs[2] and has 40 of its products listed on the VMware Compatibility Guide for vSphere 5.0. It has been particularly active over the last year and a half in addressing VMware’s vStorage APIs[3] in its software.
IBM is a VMware Technology Alliance Partner – the highest level of technical collaboration, built on a working partnership begun at the start of the last decade. Both companies are working together to improve the implementation of virtualisation at the enterprise level.
All of this may come as a bit of surprise, given IBM’s support for its own and server hypervisors and KVM, as well as VMware’s status as a software vendor eager to support all hardware environments.

IBM Storage Hypervisor – SVC Platform And TPC Management

Like VMware’s server hypervisor, which has both the vSphere virtualisation platform and vCenter virtualisation management, the IBM storage hypervisor is both a storage virtualisation platform – SVC[4] – and storage virtualisation management – TPC[5]. Both VMware’s server hypervisor and IBM’s storage hypervisor offer similar value to their respective parts of the IT infrastructure. In particular:

• Pooled physical resources from virtually any hardware vendor or tier
• Mobility of virtual resources between any physical devices
• Common capabilities regardless of hardware choice
• Centralized management

Storage hypervisors’ relative youth make them subservient to server hypervisors, with each one needing to be tuned to specific environments. IBM is also actively involved with addressing KVM as well, while Virsto has separate versions of its storage hypervisor to Hyper-V and KVM in server and VDI implementations.

IBM Storage Hypervisor Integration With VMware APIs

The IBM storage hypervisor allows users to build disaster recovery and downtime scenarios without the need for identical data centre configurations or single-vendor storage hardware. It’s integrated with VMware Site Recovery Manager (SRM) offering users the potential of reducing the high costs of traditional High Availability (HA), where configurations have to be matched precisely. It should also allow smaller companies to ‘backup to the Cloud’ using Service Providers running IBM’s storage hypervisor, rather than building their own second site. In such instances IBM’s customers will be its partners, some of which will build services using IBM’s Cloud Computing reference architecture.
IBM’s storage hypervisor also integrates with a number of VMware APIs. In particular:

• VAAI[6] – offloads functions to the storage hypervisor, improving performance as a result
• VADP[7] – provides data protection for snapshot backups at the VMware- rather than at the LUN-level, allowing users to concentrate on the value of the virtual machine itself, rather than the physical location of its associated data
• VASA[8] – enables storage tiering and allocation and a clear picture of how to locate various VMs

At the management level IBM also allows the management of storage hypervisor components as plug-ins into VMware vCenter.

Stretching Clusters Beyond 10km to 300km

The speed of light acts as a barrier to synchronous computing, because beyond 100km high latency reduces activity below acceptable levels. IBM has sold its storage virtualisation platform, SVC, in a Stretched Cluster to over 200 Enterprise customers to date, integrating with vMotion to allow virtual servers and their storage to be moved transparently across data centres (Figure 1) and easing the current processes in which VM and storage issues are dealt with separately. In its latest software release IBM has also extended the maximum distance from 10km to 300km. This should make the software more attractive to very large organisations.

Future Collaboration Around Storage Hypervisor Concepts

In term of near-term releases IBM hopes to certify its software for VMware’s Metro Storage Clustering (vMSC), allowing live-migration of VMs[9] over short distances, while expanding the number of its offerings working over ‘geographic’ vMotion distances of 300km. It is also fitting IBM technology to VMware’s Pluggable Storage Architecture (PSA), which is a multipathing programme allowing algorithms included on the server hypervisor to be offloaded. You should also expect to see a number of custom plug-ins for other IBM software products into vCenter and vSphere.
Hints of IBM’s intentions for the medium and long term were demonstrated at VMworld in Copenhagen and Las Vegas in 2011 where it was one of 5 partners demonstrating integration with VVOL[10]. It is also working on VASA API integration, which will allow users to define and segment storage, providing automatic tiering between different levels and easing the use of Solid State Disks (SSDs), front-side caching and other new technologies.

Addressing VDI Latency And Cost Issues

In VDI[11] implementations many end-users will share similar boot images – so the thin provisioning of snap copies of common gold boot images is important for efficiency. However latency and cost are major problems, especially when thousands of users in an organisation create ‘boot storms’ by logging in around the same time. IBM supports the aim of reducing the cost per virtual desktop to around the same level as cheap physical laptops. Its Easy-Tier software (incorporated into its storage hypervisor as well as a number of its arrays) provides automatic tiering of storage between SSD, SAS[12] and near-line SAS – so popular images can be made more available than those used less regularly. In this area IBM works with VMware View Composer and claims to have reduced latency and storage capacity substantially – from 206ms to 8ms and by 60% respectively in one example. In keeping with one of the key value points of a hypervisor approach, Easy Tier in IBM’s storage hypervisor allows users to take advantage of this capability regardless of their disk array vendor choice.

Some Conclusions – A Mix Of Ecosystem And Storage Hypervisor Plays

IBM and VMware share a common approach to the importance of virtualisation, but come from different backgrounds. The former is the oldest player, having introduced virtualisation in the 1960s to allow scarce computing resources to be shared; it’s a full-range hardware, software and services player and has probably forgotten more about virtualisation than all the new suppliers have learnt. VMware on the other hand is relatively young – addressing the x86 server market since 2003, adding its hypervisors to allow multiple VMs to run on a single physical machine, reducing user spend on hardware significantly and successfully expanding into systems management; it has succeeded in creating an important ecosystem and has a software-only approach.
Both companies have strong technical resources and the ability to design working products, which improve performance and reduce costs for users who decide to implement advanced virtualisation: however it’s important to differentiate IBM’s activities as an Independent Hardware Vendor (IHV) from its role as a systems (and particularly storage hypervisor) software supplier. While it undoubtedly has a number of advantages in the way its own arrays work within the VMware ecosystem, its software approach is more important – especially in addressing heterogeneous hardware and in reducing the complexity and cost of large scale virtualisation for customers. SVC, TPC and Easy-Tier combined with VASA, vMotion and other APIs within vCenter could form the basis of a joint storage hypervisor approach, although the two companies would probably need to form a different go-to-market approach from their hardware-software partnership. This is a major challenge for both companies, not least because EMC owns most of VMware and both companies are already part of VCE along with Cisco.