Bladewatch.com

I was having a chat with one of the Heads of IT over at one of the multi-nationals the other day asking him if money comes into play when buying servers/network switches, and this is what he said. I’ve removed any references to company names etc, but the content remains the same:

“It depends on what we are buying and in what scale. The first thing that matters to me is value for money or at the very least the perception of value for money. Only last week we had one vendor come in and do a presentation and everything was wonderful, she was offering twenty percent of the retail price and some free bits to encourage the sale, but its only when we asked about software deployment licenses, the extra bits we needed that we found that the twenty percent was not necessarily the best deal, so we ended up buying more expensive blades because everything was included. It’s a marginal cost thing, yes I (we) want the best deal, but I want crucially the best business value, it makes little sense to save money in one part of the transaction which is then used to pay for another set of features and end up spending more.

The key thing is solving my business needs. I have constant barrage of well meaning people from inside the company saying we need to do it faster, we need to do it cheaper or more effciently and obviously I am all for that, but those are issues with deployment, with builds and dynamic computing with the future. My favourite one is the statement “.. can rebuild a server in 45 minutes”, “Fine, is the server being rebuilt on the same network, does it need any additional storage allocated, does the 45 minutes include the post build and application load and as part of that who’s managing all the licensing and layered product configuration within that 45 minutes?”. It’s so easy to quote figures without ensuring that you are comparing the same thing, besides often you spend too much focussing on the little stuff when you should be looking at the big picture.

Right now, I need to be able to deploy servers more easily, I need to be able to allocate storage, and networks in a way that does not mean I have three guys a raft of paperwork debate and when can we do this. This needs not only a new server, better switches and storage, it requires best practice and knowledge, that is what I want help with. I’m not after state secrets, but we’re currently deploying our operating system using Altiris images, we manually change networks and provision storage, who can help me reduce the deployment time, remove the nuts and bolts type stuff so to speak, but at a cost/benefit ratio that I can afford or put forward. Remember it is the business units that sign off the investment, they want results as much as I do.”

A strange thing has been happening in the wintel world for at least a few colleagues working across our sunny City of London. It is the blurring of the lines of responsibilities. A colleague (Anthony) called me up, he has a contract working for a financial in West London.  “You ever been asked to rack a server?”  No I replied, I’ve unracked servers, not racked them, I’m sure I could work out how to do it.”

“Yeah thats fine, but thats not what I signed up for, health and safety and all that. I don’t want to be installing servers. I’m a Windows guy. What happens if I damage my suit and no one seems to be answering whether I still have to wear a suit to work if I’m just installing servers that day. They’re also talking about hardware support, that’s not my job, I know enough but not detailed hardware stuff and I don’t want to be arguing over system boards with some vendor all day.”

“It’s the cost of doing business, as we commoditize the infrastructure your role by definition changes towards a server delivery role. They need servers racked and you are capable of doing it will be the management viewpoint and in some respects we could turn around and say the industry rate to rack servers will be a lot less than your rate, so do you want to get paid well for racking servers and doing the Windows stuff, if not there will be other people.”

So Anthony will have to see what he thinks. Interestingly it was the little things that were upsetting him what he had to wear, that it wasn’t his job. I wonder if a little dialogue might not have made a world of difference, it’s all about how the message is put across – “from next week you’re doing hardware swaps and racking”, away from the “guys can you help us out racking servers and doing hardware support?”

As we change the nature of IT, remove the cost, improve the delivery, I wonder what this means for roles going forward and how we manage the need to reduce cost and deliver, with the need to incentivise our teams.

Business Wire

PISCATAWAY, N.J.–(BUSINESS WIRE)–IEEE 802.3™ Ethernet protocols with operating speeds of 40 Gb/s and 100 Gb/s are one step closer to reality. Project completion and final approval as a standard are expected in June 2010.

On Friday November 20th, the IEEE 802 Executive Committee approved forwarding the draft of the next higher speed Ethernet standard for Sponsor balloting, the final of two stages of balloting. The sponsor balloting phase will commence in November. “Once the Sponsor ballot has been completed, the draft standard will be submitted for approval by the IEEE-SA Standards Board as an IEEE standard,” says John D’Ambrosia, Chair of the IEEE P802.3ba Task Force.

IEEE P802.3ba™ will be known by its full name of “IEEE Standard for Information Technology – Telecommunications and Information Exchange Between Systems – Local and Metropolitan Area Networks – Specific Requirements Part 3: Carrier Sense Multiple Access with Collision Detection (CSMA/CD) Access Method and Physical Layer Specifications – Amendment: Media Access Control Parameters, Physical Layers and Management Parameters for 40 Gb/s and 100 Gb/s Operation.”

The project aims to extend the existing IEEE 802.3™ Ethernet protocol to operating speeds of 40 Gb/s and 100 Gb/s in order to provide a significant increase in bandwidth while maintaining maximum compatibility with the installed base of IEEE 802.3 interfaces, previous investment in research and development, and principles of network operation and management. The project is to provide for the interconnection of equipment satisfying the distance requirements identified for network aggregation and computing applications.

Wow how cool would it be to scale up Ethernet to 40GB/100GB? This would further the possibilities of bandwidth both for rich media content distribution, for backups and data transfer, but also most importantly create further opportunities and possibilities in data center virtualization. The concept in which I can fail not just workloads, but data centers between geographical locations where the data center becomes the application, in which I have my applications my services running wherever the electricity is cheapest, wherever the business mandates it for operational or legislative reasons. Business continuity becomes a reality rather than an abstract what if concept, a process in which I can have New York hosting London workloads whilst London gets that power upgrade, that data center refresh or hardware refresh for energy efficiency, performance or operational reasons. Upgrading the server, the network switch during the business day, having a dynamic infrastructure with the ability to scale becomes a reality, with that come the business possibilities. Want to start a new trading floor, a new business in Namibia, a web presence specifically for the Korean market, not a problem, we have spare capacity in New York for one part and capacity in Hong Kong but we can build it here then send it over.

Business Wire

DUBLIN–(BUSINESS WIRE)–Research and Markets (http://www.researchandmarkets.com/research/1ae19c/4q09_south_korea_m) has announced the addition of IE Market Research Corp.’s new report “4Q09 South Korea Mobile Operator Forecast, 2009 – 2013″ to their offering.

Mobile Operator Forecast on South Korea provides over 50 operational and financial metrics for the South Korean wireless market and is one of the best forecasts in the industry. The publisher provides five-year forecasts at the operator level going out to 2013. They also provide quarterly historical and forecast data starting in 1Q2003 and ending in 2Q2011. Operators covered for South Korea include: SK Telecom Co., Ltd., Korea Telecom Freetel (KTF), and LG Telecom Ltd. The publisher’s Mobile Operator Forecasts are updated quarterly and are available for one-time delivery or through regular updates.

Global Mobile Operator Forecast covers 50 operational metrics of 200+ mobile operators in 50+ countries, making up 80% of the world’s population. The publisher’s forecasts are based on their proprietary, country-specific forecasting models. These models deploy multiple regression analysis and cross-impact matrices that estimate relationships between subscriber data, technology use and deployment data, overall economic and demographic changes expected in a particular country; and relate these to company operational and financial metrics.

I blogged this because it highlights the possibilities for not only mobile phones, but all the related infrastructure and services around that. As we bring consumers online to mobile technology, can we look at internet enabled hand sets, bringing 3G internet to new markets creating new opportunities and new ways of doing business or using that technology. Related to this with new markets, new user groups, would they need a pc or a thin client using virtualization on any operating system in order to get them online?

Could countries like Korea be the new markets of opportunity for infrastructure, both in terms of a server, or even moving to the next generation data center using virtualization technologies?

For the second time, I had a conversation with a colleague. They had just moved from one vendors server to another. The specifications, the underlying kit was more or less the same, a few Xeon or AMD processors, an array controller and a network card.

Janet had been given this new vendors software and been told fix it. “Make it work with our build”. The following interview happened over the phone and had no specific structure or guidance. To avoid any statements or emotional discussions I have removed the references to the vendors, but the content, the underlying message remains the same.

So what do you and what are you up to?

I work in the Windows design team, we provide the ongoing hardware validation and integration services. It sounds more complicated than it is. We validate the servers against the existing enterprise build. This includes establishing what drivers and components are needed, what components we will need to include in the next release of the build so that the engineers can deploy the new operating system to that model of server.

At the moment we are working on adapting the build for the new servers we are buying which are from a different vendor, it can be quite frustrating and surprisingly time consuming.

You say time frustrating or consuming why?

Every build version has to be approved by the different sponsors and include the necessary hot fixes, driver packs which also have to be backward compliant with the older systems that are still supported under that build. This becomes even more so when you change server vendors as we need to validate what is required, what is extra and what we can get away without installing. The fewer bits we install, the less we have to manage and update with each release or update to the build, the does the network card driver work with the network card management software issue.

So name two features of the vendors software/drivers web site that cause you pain

First one would be the web sites can often be a bit high maintenance, designed from a driver publishing format, not a server or end user viewpoint. A perfect example (I will seem a bit emotional, but anyway), if I want the driver for a network card in server three, I might have to go through different parts of the web site, I might have to register, I might need to even know the serial number/derivative or get the list of every possible component for that series of server.

Second would be the differences between how the drivers, management software and firmware all interact. The main vendors we deal with all seem to have different ways of doing things which is fine, but this simply adds to the overhead, the learning experience.

I will continue by saying thirdly (which you never asked for) the complexity in the difference between what is needed to make the server operational, what is core to systems management and what is a nice to have. There does not seem to be a matrix in terms of core that is needed to make it boot, necessary like the server agents, and nice to have like online diagnostics.

We need to make the drivers easier to upgrade in terms of a software pack, remove the end user from individual component and management updates, the more separate bits there are to manage, the higher our costs, the more we remove from the build, simply saying “it is not core or supported by our build”.

Can you give me an example?
My favourite example was the guy that upgraded the drivers on his server as part of testing only to find that he could not use the management server because the driver was not supported with that management pack, the result being that he could not easily change the network card speed using the software, and had to access it using the device manager. That caused us endless calls and comments.

Do you have a preferred vendor?
Yes but I wonder how much of that is because we are used to, how much is because we got told from now you buy these servers. Interestingly though I wonder how much of our views are because no one vendor has sat us a team down and translated features and concepts. By that I mean, no one has said you are used to so and so, this is how it works with us.

http://www.hpcwire.com/blogs/IBM-Cuts-Cell-Loose-71994607.html

A few weeks ago, I speculated that IBM would probably not develop a follow-on Cell processor for the HPC market beyond its current PowerXCell 8i. A recent report by German publication Heise Online (English translation here) seems to bear this out. The article asserts that IBM’s plans for its next-generation Cell processor have been scrapped.

According to Heise, IBM VP of Deep Computing David Turek confirmed that there will be no successor to the PowerXCell 8i, the high performance Cell variant IBM developed for the Roadrunner supercomputer and its QS22 blades. Mercury Computer Systems, Fixstars, and Sony, also incorporated the processor into a few systems, but the technology was never widely deployed. At one time IBM had talked about expanding the Cell’s Synergistic Processing Elements (SPE) from 8 on the PowerXCell 8i up to 32 on the next-generation chip. That would have kept the processor on pace, performance-wise, with the latest and greatest GPUs from AMD and NVIDIA. Apparently though, this is not to be.

There will be continued discussion about this. I remain a fan of the IBM Cell processor technology, they are meant to be great in high performance computing, but as with any processor technology the performance depended on the application being written and optimized for the platform. I’m off to read up more about the Cell processor and see if I can find out more.

PR Newswire

ARMONK, N.Y., Nov. 24 /PRNewswire-FirstCall/ — First National Bank of Namibia Ltd has selected IBM ( IBM) to help the bank localize its banking systems and operations. Under the $5 million agreement, IBM will provide FNB with two of the latest IBM System z10 Business Class mainframes and supporting software to meet regulatory requirements and support the bank’s growth.

The deal marks the first deployment of mainframe technology in Namibia. Implementation of the IBM system was completed this month and forms part of FNB Namibia’s $15 million project to localize its core banking systems and operations in Namibia.

Previously, FNB Namibia, which is majority owned by FNB South Africa, had its server infrastructure and the data of more than 400,000 customers residing in South Africa.

An article to remind us of the opportunities in emerging economies both in terms of services and products. It’s exciting to here what’s going on from a technical and business viewpoint, for example as we onboard businesses and users in China, what possibilities for revenue, for product development and innovation will we see going forward?

Building

Dumfries and Galloway planning officials have given the go ahead to UK’s largest data centre, set to be built in Lockerbie.

The £950m, 250,000m² project is being developed by Lockerbie Data Centres, a subsidiary of local firm, Robison and Davison, and will be attached to a business park with 18,000m² of high-tech office space.

The development will create 3,000 jobs in construction over a five to 10 year construction period – 900 through business attracted to the business park and 50 to operate the data centre. It will attract an estimated £3.5bn of inward investment, according to the firm.

Very cool, I wonder what technologies they will be using as part of the data center design in terms of energy efficiency and best practice. It will be interesting to see if the data center is being built with a specific use or market place in mind, Grid computing? Virtualization in cloud or data center space? We’ll have to see. I’m off to read up more.

Public Technology

Kings College London (KCL) has unveiled a new IT infrastructure that will improve service availability, security and user access for its 19,700 students and 5,000 staff.

Implemented by Getronics, KCL’s new ‘Global Desktop’ uses desktop virtualisation technology to simplify and optimise the way IT services and applications are delivered. Based on Getronics’ ‘Your Workspace, Anywhere’ initiative, it is designed to reduce operating costs and streamline processes.

In a five-year contract, Getronics is working in close partnership with the College’s Information Services & Systems (ISS) department to deliver their vision of a virtualised environment running on thin client technology. The virtualised desktop environment meets the objectives set out in the College’s Ten Year Strategic Plan (2006-16) to enhance quality and distinction in all areas of business activity, supported by world-class professional services and infrastructure. The project will bring a broad set of features and functionality to a diverse set of users.

A great example of how this college managed to improve delivery and availability using virtualized desktop and thin client technology. It is amazing to see what can be achieved and what opportunities created with the right set of technologies to deliver an adaptive infrastructure. I’m off to check it out.

PR Newswire

ARMONK, N.Y., Nov. 19 /PRNewswire-FirstCall/ — IBM ( IBM) supercomputers are the most energy efficient in the world, according to the latest Supercomputing ‘Green500 List’ announced today by Green500.org.

The list shows that 18 of the Top 20 most energy efficient supercomputers in the world are built on IBM high-performance computing technology. The list includes supercomputers from Saudi Arabia to Germany and the United States that are being used for a variety of applications such as astronomy, climate prediction and pharmaceutical research. IBM also holds 69 of the Top 100 positions on this list.

Energy efficiency — including performance per watt for the most computationally demanding workloads — is a core design principle in developing IBM systems. IBM offers the broadest range of generally applicable supercomputers represented on the Green500 List including Blue Gene, Power servers, iDataPlex, BladeCenter and hybrid clusters.

Check out this article talking about IBM’s ranking in the Supercomputing Green500 List, it’s great to see these kind of announcements which illustrate the possibilities of Super Computing and how with the right range of technologies and configuration, we can still deliver high performance solutions in an efficient way gaining a high performance per watt. Do check it out, Dell, HP and Cray are on the list (to name a few).