Converged Infrastructure - Hyped!!
There has been a lot of hype about converged infrastructure; the largest growth area in datacentre technology post 2011 has been converged infrastructure (CI). To set the stage, let’s start looking into the definition of converged Infrastructure. Simply said, it is a configuration that combines physical compute, storage and network validated for specific workload to form an “out-of-box” solution offering management through single console, ease of patching – one patch set for the box, single point of contact for all support – single throat to choke and ease of deployment – ready when arrives.
Though if not carefully considered, the pros are outweighed by its constraints, to keep the pros in play you have to color between the lines and that too with specific approved pencil.
These systems are designed for simplified user experience, allowing IT generalists to provision, operate and support against requiring an administrator with special skills. SMBs could realise big benefits from CI, who wants an in-house team to focus on the users and applications that runs the business. An organization achieves operational simplicity but it sacrifices configuration flexibility. Though for “startups”, CI could prove an overkill solution considering the size of organization and operations.
The concept of hyper convergence emerged from CI providing similar benefits but addressed shortcomings of converged infrastructure like rigid configurations – resource ratios are fixed, leading to waste of resources due to saturation and blocking granular scalability. Also the upfront capacity planning is important for CI which goes against the more elastic and scale-out infrastructure created through addition of incremental resources. Acquisition cost of CI is higher (20 to 25 percent more) vs. buying exact configuration to meet your requirements. With single “throat to choke” support results in vendor lock-in – “all eggs in one basket” and lack of freedom to reuse or extend CI for other than specified workload.
While the former combines compute, storage and network in one validated configuration, later builds single infrastructure stack merging multiple efficient elastic pool of x86 resources allowing simplified management.
Software Defined Data Centre (SDDC) technologies became the key driver for hyper converged infrastructure. Since it is run by the software it provides the flexibility and agility demanded by the most businesses from their IT departments today.
Software being the basis of hyper convergence, it provides the flexibility required to meet current and future business needs without having to rip and replace infrastructure components. The benefits of new features can be immediately realized without having to change or replace the hardware as it only requires the upgraded software to be applied.
Software defined data centre has many characteristics including virtualisation, automation and offering IT as a service (ITaaS).
Hyper-converged infrastructure use virtual machines (VMs) as the most basic component of the environment and SDDC being the foundation of virtualization, the major OEMs are pushing this technology very hard as the revenue benefits associated with software driven technology is more lucrative in terms of higher margins and revenue growth over box built hardware infrastructure products business evolving at very slow pace. It also eliminates very common management and cost overheads associated with manufacturing, supply chain and logistics.
Finally to conclude, CI is for sure, not one-size-fits-all solution, rather with infrastructure in the box, there may be lot of throw away from existing investment which can be protected by adopting hyper converged infrastructure. It can be adopted with existing infrastructure solving your immediate challenges and producing major benefits added with simple designs, decreased administrative overhead, and simplified vendor management to highly virtualized environments ignoring some insignificant scalability overheads along with potential risk of over sighting management needs.
Considering my own datacentre deployment with more than 350 servers and 200TB of storage which is 80 percent virtualised to cater test & development workloads, some legacy production systems hosting SQL databases along with file and print applications and an ERP system on independent traditional infrastructure without any virtualization, I believe that full blown hyper convergence is yet to see the sunny days in non ITeS companies due to lack of confidence garnered by limitations of technical know-how and absence of meaningful PoC demonstrating relevant use cases. Workloads like VDI, dynamic test & development environment and disaster recovery implementations are the sure shot use cases of hyper converged infrastructure and it is proving to be most promising technology for the companies offering SaaS and cloud services.
In the end, the question is: “What delivers the best value for your particular organization?” must be answered by individuals considering key performance indicators of operational efficiency improvements, reduction in business risks and providing flexibility & agility to support changing business needs within available resources and budget constraints.