Racemi’s Image-Based Provisioning Ripe For Cloud Computing

Racemi’s Image-Based Provisioning Highlights

• Enables the migration of bootable server images between physical and virtual machines, as well as in house and Cloud Computing
• Provides heterogeneous approach help to large enterprises
• Its technology partners and OEM resellers include BMC, CA, Oracle/Sun, VMWare, Emtec and Cisco
• Competes with IBM Tivoli, Symantec Altiris and Novell Platespin
• Claims to be less complicated than script-based provisioning and cheaper than other image-based offerings
• Concentrates on heterogeneous environments with multiple operating systems and hypervisors

‘Racemi’ (‘a bunch of grapes’ in Italian) was a name picked by Charles Watt, the founder of this specialist start-up, to complement its early work on blade servers. It’s quite possible that you won’t have heard of them, as their go-to-market approach is exclusively through OEM contracts. However if you’re a CA Spectrum Automation Manager or BMC BladeLogic customer you may already taking advantage of its 10 year development of image-based provisioning – a technique for deploying multiple servers in large data centres. I recently had the opportunity to interview the company’s CEO Lawrence Guillory to discuss the company’s strategy, developments and competitors. I believe this enthusiastic small company has a vital role to play in server management and large organisational Cloud Computing.

How Racemi Became An Image-based Provisioning Specialist?

Racemi’s initial work was in designing the world’s first blade server. Blades are a small part of the market today, supplied by major systems vendors such as HP, IBM, Fujitsu and Oracle/Sun, typically to large companies. They have proprietary designs with the server ‘blade’ fitting either vertically or horizontally into a vendor-specific rack-mounted chassis. Blades enable better account control than industry standard designs, which limits customer choice: however they also allow for a tighter integration of processors, allowing for faster and deeper processing.

In 2003 having decided that the additional investment (perhaps as much as $100 million) needed to sustain itself as a blade server supplier was beyond it, instead it began to concentrate on building the software needed to manage the deployment of processing, storage and network elements within these and other new-style servers. Unlike VMware, Racemi also decided early on that it would be an OEM supplier – helping systems management suppliers with these intricate and complex challenges. Over the years it developed enough expertise to declare itself as an image-based provisioning expert. Initially these techniques were used almost exclusively in Scientific/Technical Grid computing – another small part of the server market; however the development of Cloud Computing is widening Racemi’s horizons substantially.

What Is Image-Based Provisioning?

Having deployed a server, IT technicians almost always write scripts to make the task easier for building the second and subsequent machines. Racemi went one step further by taking the image of the system (the operating system, software, drivers and settings) as the starting point for deployment. It’s DynaCenter software stores bootable server images in a library, including applications, operating systems and configurations.

Racemi has developed deep domain expertise in the specifics of how applications run on servers, both in terms of the operating system installed on the physical machine and how those are then transferred for use under hypervisors on a virtual machine. Supported operating systems include Windows, Linux, Oracle’s Solaris and IBM’s AIX. Lawrence claims that Racemi has very little call from customers to support HP-UX, but that IBM’s Unix is has a life on its big machines and that supporting Solaris is useful for those companies looking to migrate Solaris applications from Sparc onto x86-based servers. Supported hypervisors include VMware’s ESX, Microsoft’s Hyper-V, the Open Source Xen Hypervisor, IBM Logical Partitions (LPARs) and Oracle/Sun’s Logical Domains (LDOMs).

The heterogeneous nature of Racemi’s approach is very useful for large organisations that want to move their server workloads – both from physical to virtual machines and from in house to public Clouds. In particular (in reference to Figure 1):

• Cloud to Cloud (C2C) – Once deployed on a public Cloud, DynaCenter can facilitate moving to another; this should prove particularly useful for large Cloud suppliers and their customers in maximising the advantages of location or legal requirements of compliance
• Physical to Cloud (P2C) – many organisations are looking to lift non-core applications and deploy them on public Clouds; to do so they need first to move from a physical to a virtual machine deployment; Racemi’s approach allows the journey from physical to virtual to Cloud to be automated, saving many the expensive of rewriting the application
• Cloud to Virtual (C2V) – many early Cloud Computing suppliers have failed to provide their customers with a clear exit strategy; Racemi’s software allows images to be taken from public Clouds and redeployed in house on virtual machines

The development of server virtualisation and Cloud Computing is multiplying the number of times a server needs to be provisioned. Whereas in the recent past most servers might be provisioned a couple of times during its lifetime, we’re entering a stage of the market in which a single server may run a number of different images each day. This trend is likely to become even more extreme with the acceptance of offerings such as Cisco’s Unified Computing.

How Does Racemi Compare With IBM, HP And Other Vendor Approaches?

Lawrence gave his view of how his company’s approach compares with his competitors. In particular:

• HP uses script-based provisioning, which he considers to be very complicated in comparison with Racemi’s approach – a laborious old way of building servers in a kind of assembly-line fashion; he believes it is impossible to build a ‘service catalogue’ with HP’s current techniques; coincidentally he’s very much like to sign HP up as a customer
• IBM Tivoli’s approach is solid and image-based, but Lawrence considers it to be comparatively expensive and hard to understand; he claims you virtually need a doctorate to add services
• Symantec Altiris takes a windows Disaster Recovery view of the world, which is good for SMEs, but is less relevant to the heterogeneous world of server use in large enterprises; in his view it has a bare metal focus, limited automation and little image mobility
• Platespin has been a strong competitor in the disaster recovery area, however he believes it has become somewhat lost following its acquisition by Novell

Lawrence’s opinions of his competition are clear very subjective, but the acceptance of Racemi’s products by technology and OEM reseller partners like BMC, CA, Oracle/Sun, VMWare, Emtec and Cisco demonstrates the degree to which provisioning is an area big systems management companies are prepared to buy rather than build themselves.

Racemi’s Future – Building Its Own Brand

Lawrence believes that in the future Racemi will be in a position to build its own brand by keeping Racemi’s large OEM partners and while moving down market in the SME space where there is no conflict, have a light-weight product to sell via reseller channels. In particular he can see a time when the company makes a simplified version of its software available for SMEs to download, restricted to one or two operating systems and/or hypervisors.

Some Conclusions – Racemi’s Image-Based Provisioning Will Power Cloud Computing Adoption By Large Enterprises

Server management was already getting difficult before the move to industry standard virtualisation and it’s getting harder with arrival of Cloud Computing; today it is almost impossible to run large data centres without significant amounts of systems management and automation. I’ve no doubt that an important part of this is Image-Based Provisioning, which is becoming a necessary component for the adoption of Cloud Computing by large enterprises. By maintaining a library of bootable server images (including applications, operating systems and configuration settings) server managers will be better able to move applications in and out of Clouds and in between physical and virtual deployments.

Racemi has spent almost of its life focusing on this narrow specialisation; in fact its future would be far less bright if these techniques (originally associated with Scientific/technical Grids) were not now becoming key in the commercial market. I’ll be watching the company, its OEM, resellers and competitors closely over coming months.

Do you use Racemi software? Are you still using script-based techniques for provisioning? Please let me know you experiences by commenting on this article.