• United States

How does utility computing keep up with shops that have frequent demands?

Dec 15, 20054 mins
Data Center

* On-demand for businesses whose demands shift on a more frequent basis

Last time we spoke about applying the utility computing model to IT shops that support seasonal businesses. A more complex set of capabilities is needed when applying this model to businesses whose demands shift on a more frequent basis. 

How, for example, could a company ever hope to manage the allocation and reallocation of resources if those shifts happened on a daily basis? The operational expenses involved in continual reprovisioning would destroy most IT budgets.

The answer here is of course to automate all provisioning services that lie at the heart of the on-demand model, and by so doing, removing the need for human involvement and eliminating the major cause of unplanned downtime – human error. How close we are to providing automated management of the provisioning function might surprise you.

CA, EMC, HP, IBM and Sun have all made good progress when it comes to asset discovery at most points on the storage topology (a big thanks to the SMI-S standard, the enabling technology for storage-area networking-based storage – and hopefully soon, for network-attached storage-based storage, as well). The first four companies have also made good progress in automating the provisioning process (this could also be the case with Sun as well, but I haven’t seen that yet).

How does SMI-S provide foundation technology for on-demand environments? Broadly speaking, it works like this.

Asset discovery is the first step in automating the provisioning process. Understanding the availability of storage devices – disks, tape libraries, switches, HBAs and the fabrics to which they are all attached – enables monitoring; monitoring enables reporting; reporting enables analysis; and analysis – in most respects the toughest part – looks at what is going on and provides the predictive model on which proactive management can be based.

The analysis piece of course is the “super sauce.” The analytics read the incoming report data to do event correlation so that it understands (and in some cases, to pre-empt) problems. The information is then passed upstream to a policy engine that can look at available assets, interpret what must be done according to a set of user-defined policies, and then pass the correct decisions up to the management interface that will manipulate the environment. 

Ideally, in the course of all this interpretation, analysis and decision-making, everything is measured against a well-defined set of service levels. In this way, policy-based services can be optimally matched up with the service levels defined in those service-level agreements.

When all of this happens in an on-demand environment, assets can be brought online and released back to the virtual asset pool according to whatever the defined service levels are. Assets are provisioned, data is moved, and assets are reassigned, all according to the needs of the user base. Link all this with a charge-back application and even the accountants can be happy campers.

* * *

And now, the answers to last week’s quiz. For those of you who missed it (at this point those people should begin to do their ritual Dance Of Shame), the question was to identify the vendors that are marketing their automated utility environments using the terms that follow. The terms and the companies using them appear below.

1. Adaptive Enterprise (HP)

2. Autonomic Computing (IBM)

3. Darwinian Architecture (HP)

4. N1 (Sun)

5. On demand (IBM)

6. On-demand (EMC, and just about everybody else in the world)


If the topic of automation interests you, you will want to participate in Enterprise Management Associates’ periodic review of automated system and storage management services.  We welcome IT personnel to participate in this research. To access the 10-minute Web survey, please click on either of the two links appearing below.  Participants will receive a copy of the final report.

For the “Automated Storage Management” survey:

For the “Automated Systems Management” survey: