Physical Resource Management Holds the Key to IaaS
Enterprises ignore it at their peril.
Looking at the way that cloudspeak has permeated just about every IT discussion I hear today, it nags at me that certain topics have been going mostly unaddressed -- like the need for physical resource management to make all that virtualized and software-defined stuff work at all. Until recently, nobody has really been talking about it.
I first thought about this a couple of years ago, when the National Institute of Standards and Technology (NIST) published a paper on Infrastructure-as-a-Service (IaaS). At the time, I viewed it as a savvy, if subtle, tome that decimated the story coming from certain hypervisor vendors about IaaS, the new and improved data center solution enabled by server virtualization, software-defined networking and software-defined storage.
NIST noted that these "enabling technologies," which were then (and are today) in differing stages of standards coalescence, each fraught with their own technological hurdles and vendor in-fighting, did not comprise in and of themselves any sort of true IaaS model at all. IaaS, NIST observed, requires three things to become reality.
First: orchestration and administration tools capable of working across all the diverse interpretations made of hypervisor computing and software-defined infrastructure. To deploy a service, you need to have some method orchestrating the deployment of different resources from resource pools, in the right quantity and configuration, to meet the application need. Also needed are administration tools that enable management of that infrastructure over time, constantly adjusting to workload requirements.
A Layered Approach
The second IaaS requirement calls for a bunch of operational activities comprising what the organization calls a "management layer" and an "operations layer." The management layer consists of a bureaucracy created for managing service delivery and configurations, and ancillary services like data protection and security, across the entire infrastructure.
The operations layer is another bureaucracy for managing change and handling things like request fulfilment. This is the basic work of the staff of a conventional datacenter today, comprising both the labor cost component of the datacenter budget and accounting for many of the impediments to realizing agility goals.
The third requirement for IaaS is a service delivery layer, composed of activities for operating IaaS as a business. Here, NIST lists activities like financial management, demand management, business relationship management, service lifecycle management and so on; basically the business management activities to which every successful service provider must attend.
Poking VMware in the Eye
The NIST document struck me as a big red hot poker in the eye of VMware and some other vendors who claim that they're ready to displace 40 or so years of datacenter architecture and operations with flimsy and half-baked hypervisor technology mixed with a proprietary stack of software-defined technologies and cloud woo. I confess to experiencing a bit of schadenfreude
at the time, as I turned the pages of the NIST document.
Of course, who listens to NIST? Things proceeded to develop around the idea of Software-Defined Datacenter (SDDC), a reworking of IaaS introduced a year later because, well, trendy terms have a shelf life. No one really seemed to recognize the truth exposed by NIST: that the new-fangled datacenter would, out of necessity, need to have attributes that would make it look (and cost) an awful lot like the original datacenter. Perhaps even more importantly, work on "software-defined" proceeded mostly apace with a disregard of the need for universal resource management.
What is resource management? It is the management of the hardware componentry and interconnects that underlay all of that sexy virtualization of servers, storage and networks in the Brave New World of SDDC. Nobody likes to think about resource management today, any more than luxury cruise line passengers at the beginning of the 20th Century liked to think about the coal tenders working in the bowels of their ships. Without the constant feeding of fuel to boilers, and the maintenance of the steam engine components, the ship would founder.
So it is with resource management. Without it, you can separate all of the control path from the data path, all of the software services from the hardware services, all of the logical from the physical; and your cloud or SDDC or IaaS will still be junk.
The good news is that I am now seeing a small and growing group of vendors who have begun including the need for hardware management into their software-defined product pitches. FalconStor's FreeStor analytics are an example of what can be done, as are IBM's latest innovations around Spectrum Storage Analytics. Even ioFabrics and Cohesity have added hefty doses of physical resource monitoring and management into their pitch for software-defined storage infrastructure.
Nice to see a few vendors getting real about the less glamorous requirements for realizing an SDDC vision.
Jon Toigo is a 30-year veteran of IT, and the Managing Partner of Toigo Partners International, an IT industry watchdog and consumer advocacy. He is also the chairman of the Data Management Institute, which focuses on the development of data management as a professional discipline. Toigo has written 15 books on business and IT and published more than 3,000 articles in the technology trade press. He is currently working on several book projects, including The Infrastruggle (for which this blog is named) which he is developing as a blook.