Earlier this morning, Intel has confirmed that it is discontinuing the development and support of its NAS Performance Toolkit, an application that measures the overall performance of small network-attached storage systems.
However, the software will live on, as Intel has released more source code for the tool. The NASPT, as Intel likes to call it, isn't just simply the kind of tool one would use to do serious testing of an enterprise-scale NAS, as it lacks the ability to test storage devices under specific application loads-- something that comes way too short for the average network manager.
But the software was a bit more useful for tests of NAS servers intended for use in smaller-scale environments, thanks to its inclusion of tests for performance when used to open and close files from personal productivity applications.
And simulating data backups was another of the application’s overall capabilities, as was testing a NAS’ abilities to cope with video playback and recording.
Those features made the free software a useful item in system administrators’ resources when planning and implementing small networks in which a NAS device substituted for a server. That’s a role for which small NAS systems, now often equipped with four or more drives and gigabit Ethernet, are very well-suited for.
System builders using open source tools like FreeNAS or Openfiler also often used the tool to test their newly-built storage appliances, while hardware reviewers were among its other early adopters in a way.
But it’s rather difficult to know just when Intel decided to discontinue the software, as a document.last.modified query on many of Intel's web pages often returns the day of the query as the answer.
It’s also difficult to understand why-- the company last year renewed its push for Atom-powered NAS appliances aimed at small businesses.
Whatever the reason for NASPT‘s demise as a supported product, Intel has been good enough to release the source code for the tool as a Visual Studio 2008 project. It will nevertheless be interesting to see in the next coming months how the chip giant will manage this.
In other IT news
Oracle said today that it's rolling out some of its solutions as service clouds, with capacity-on-demand (CoD) pricing, based on its various engineered systems configurations.
Financial services operations have stringent and significantly higher capital requirements since 2008's financial meltdown. Today, every single dollar counts and any shift from the capital expense of the ledger to the operating expenses is welcomed, says Juan Loaiza, senior vice president of systems technology at Oracle.
And many companies in other industry segments are equally keen to conserve cash for rainy days or acquisitions, and are therefore very cautious of big capital investments in information technology, at least if they can avoid it, that is.
That's why Oracle is taking a trip back to the tail-end of the dot-com boom and rolling out a new capacity-on-demand pricing program for its 'engineered systems' coupled with a monthly cloud pricing program that allows users to enjoy Oracle gear installed in their very own data center while also enjoying economics that are similar to what Oracle will eventually provide as raw infrastructure services out on the Cloud.
Oracle announced it was going to do infrastructure cloud services in October 2012, but it has not put the raw compute capacity into preview yet. Raw storage is in tech preview, however.
Oracle gave a preview of the private cloud variants of its engineered systems at the first Cloud World event in Dubai last week. The rest of the world will get to hear all about it in an online event hosted by co-president Mark Hurd and Loaiza given tomorrow.
Loaiza, who has been at Oracle since graduating with his master's degree in computer science from MIT in 1988, is in charge of the mission-critical features of the Oracle database and is also in charge of the Exadata clusters.
The engineered systems that now come with cloud pricing include the Exadata database clusters, the Exalogic middleware clusters, and the Exalytics in-memory database appliances, which are all based on Intel's Xeon processors.
The Sparc Super Clusters, based on Oracle's Sparc T4 processors and Exadata storage servers, are also engineered systems and are also available with the special packaging and pricing. So are the Sun ZFS Storage Appliance arrays because sometimes you need external storage with brains in addition to what is actually embedded in these preconfigured clusters.
"We're trying to bring public cloud benefits to a private cloud," Loaiza said. "This is designed to be a very financially attractive model, and it's a good idea. It's in fact a win-win for all concerned."
But the private cloud deal is actually very straightforward. You figure out the configuration of the server (s) that will meet your peak performance needs. Perhaps you will need to have excess capacity during the end-of-year shopping season, or even once a quarter, it all depends on your specific needs.
Next, you configure an engineered server with about 75 percent of that capacity and you configure it with Premier Support, and the new Platinum Plus Services, which is the quarterly proactive analysis and advisory services.
The Platinum Plus services have terms of services that are more attractive than Premier Support, including five minute fault identification, fifteen minute system rebooting, and faster escalation to Oracle development if there is an issue with the system
Oracle does all the security patching on the system as well, since it knows what is on the server and configured it itself.
The company bundles this all up and sends you a server with 75 percent of the compute capacity activated in the engineered system. All of the other features of the systems-– memory, disk, and I/O –re fully configured on the server.
To take part in the private IaaS cloud deal, you have to commit to having the server on site for three years and to calculate the monthly fee that Oracle will charge you, you then divide the price by 36. Oracle isn't charging any interest, setup fees, or fees to cover the value of the latent CPU capacity shipped in the server.
You can then activate that 25 percent latent capacity, and the monthly fee is directly proportional to what you are paying Oracle for the CPU capacity in the engineered system. For example, if you paid $750,000 for an engineered system plus support, you would pay Oracle $16,666 per month.
If you wanted to activate the latent capacity for a month, you would pay an extra $5,555 for that month and then turn the capacity off when you aren't using it.
