Bladewatch.com

Dell community

Dell Point of View:

• Data deduplication is getting significant attention from a growing number of customers looking to manage ever-increasing storage growth by eliminating redundant data. Dell believes data deduplication is just one – not the only – option to help customers meet this challenge.

• Dell is taking a customer-specific approach in which Dell consultants will guide customers in determining how any or all of the following optimization considerations can help them manage their storage as efficiently as possible:

o Thin provisioning
o Reclaimed storage
o Storage virtualization
o Automated load balancing
o Data tiering and classification
o Deduplication

• Dell’s deduplication strategy is to foster and encourage the rapid evolution of dedupe technology into a storage environment where the functionality exists everywhere. As deduplication matures quickly, it will move beyond backup storage – where it primarily resides today — to other data types including near-primary, archive, file and object storage solutions.

• Dell believes that incorporating deduplication functionality into ISV, application and storage software can provide significant benefits to customers including:

o More affordable, better performing solutions than purpose-built appliances.
o Simplified management through single console interfaces for data protection and management.
o Reduced network traffic across WAN and LAN infrastructures leading to improved, more cost effective disaster recovery environments.

I was on the Dell comunity blog the other day looking at what’s new, and noticed this article about data de-duplication, something that I think is getting more discussion in the enterprise whether it’s to manage our storage requirements, or be more efficient with the storage we have. Anything Dell or the other vendors can do to help with this has to be a good thing, we do need to look at the data we have, how we provision and manage it, simple steps can transform your storage requirements, user profiles for example, do they need backed up or do they need backed up and replicated?

Share and Enjoy

HP

L’ISLE d’ABEAU, France, June 18, 2009 – HP today announced the availability of standardized assessments from HP Critical Facilities Services in Europe, Middle East and Africa (EMEA).

In this difficult economic environment, large capital projects are being put on the back burner as customers face challenging demands. Data center downtime, capacity constraints and increased operating costs directly impact revenue, as well as customers’ ability to support business growth.

To tackle these challenges, HP Critical Facilities Services now offer new standardized assessment services that enable companies to improve efficiency and lower costs, without having to wait for large budget approvals. These services include:

· HP Energy Efficiency Analysis helps customers improve the energy efficiency of their data center. This comprehensive study benchmarks a customer’s data center against industry best practices for power usage and infrastructure efficiency. It provides scenarios and cost benefits along with detailed plans for energy-efficiency improvements, investment payback and facility reliability.

· HP Basic Capacity Survey provides quantitative capacity and availability information to help customers more effectively prioritize and allocate infrastructure resources.

· HP Infrastructure Condition and Capacity Analysis enables customers to improve the reliability of their facility by providing a system-by-system evaluation of their critical technologies. It also evaluates the power and cooling capacity of the facility infrastructure in comparison with current power and cooling loads, providing customers with quantitative information for capacity planning or future expansion.

· HP Comprehensive Assessment allows customers to optimize data center cooling to reduce operating costs, boost capacity, and improve reliability. Comprehensive Assessment utilizes Computational Fluid Dynamics (CFD) modeling of the entire supply and return air flow to provide recommendations for improving cooling effectiveness.

These cost-effective assessments identify ways to immediately increase capacity, lower operating costs, and increase availability.

“Many customers are facing challenges as a result of capital budgets and lower operating costs,” said Ed Ansett, director of critical facility services, Technology Services – EMEA, HP. “This suite of standardized assessments provides a comprehensive framework and roadmap to increase efficiency and generate increased economic returns.”

HP Standardized Assessment Services are now available in EMEA as pre-packaged and pre-priced offerings.

More information about HP Critical Facility Services is available at www.hp.com/go/eypmcf.

Anything the different vendors can do to aid end users with their capacity planning and in their data center management has to be a good thing, I’m off to read up more. The more we discuss the data center and couple it with application capacity planning, the more we can establish or give a value to the application cost in watt/space terms to help understand what applications we want to host, and which we might buy in or operate on an enterprise cloud solution. Could we not buy in data center space or capacity for our grid applications and use our data center for our in-house trading applications and back office functions?

Share and Enjoy

PRNewswire

TROY, Mich., June 15 /PRNewswire/ — Altair Engineering, a global provider of technology and services empowering client innovation and decision-making, today announced that the U.S. Department of Defense has exercised its option to continue using Altair’s PBS Professional software as the standard workload management solution for its High-Performance Computing Modernization Program (HPCMP).

The Altair software currently schedules runs on the department’s centers for scientific research and development and for testing and evaluation. These high-performance computing centers include facilities at the Aberdeen (Md.) Proving Ground; Wright-Patterson Air Force Base near Dayton, Ohio; and the Navy DoD Supercomputing Resource Center on the Mississippi Gulf Coast, as well as sites in Vicksburg, Miss.; and Fairbanks, Alaska. In all, the systems at these facilities furnish a total 1.3 petaFLOPS of computational power (1.3 thousand trillion floating point operations per second).

In 2006, Altair earned a two-year contract to provide PBS Professional for these centers, with a series of eight one-year renewal options. The HPCMP now has exercised the Year Three option, based on the performance of the software and the flexibility of Altair Engineering in adapting to the needs of the organization’s constantly changing environment.

“We focus on making the most secure, reliable, and scalable software in the industry,” said Dr. Bill Nitzberg, PBS GridWorks Chief Technical Officer at Altair, “and we’re very pleased that the Department of Defense continues to trust PBS Professional to manage some of the most sophisticated machines in the world.”

Check out this article illustrating how the Department of Defense are using Altair as part of their HPC solution, it’s always interesting to see what technologies are being used and how they are delivering operational or business benefits. I’m off to read up more.

Share and Enjoy

I’ve been reading a lot about data center efficiency and speaking with data center managers, in fact I had one telling me over the phone about how project managers, “just don’t understand the power and space constraints we have, that we’re asked to install more in the data center but still maintain resilience..” the magic N+ value. Interestingly the issue is not the data center when you ask for more information about how IT works, what the supply conditions are, how the billing and process works, about how we provision and deploy servers.  As we move towards virtualization this is less so as we move to commoditized platforms which deliver virtual machines to the end user.

Anyway, I asked the following:

• Who defines which servers you buy – the client does with the project manager and the unix/windows boys
• How do you charge to host a server  – oh it’s a fixed cost per device/server it’s £450 a year
• How defines when a server is decommissioned – the customer logs a request to decommission a server when they are ready
• What is your oldest x86 device – A Compaq Proliant 4500
• How much virtualization do you have – we have it, but it’s not been signed off for use by the business, so only IT use it.

So if we look at these key questions and answers above, the barriers to success are not in the data center, they are in the operations. There is no doubt the data center might be better configured, it might be optimized for air flow, have more power or cooling installed, but that wont fix the actual process and policy issues and regardless of how much you invest in your data center, if you don’t start with the demand/supply issues, you’re going to continue investing or eating data centers.

Key steps to take:

• Define rules for server age – any server over 4 years old to be decommissioned/virtualized or migrated to new hardware – there will be exceptions to the rule based on business/application compability of course, but the commodity stuff should be replaced with more efficient systems.
• Charge per watt and space in the data center – you want to install a 12u bad boy server that needs many megawatt that’s fine, you pay more than Mike who needs a DL380 G6.
• IT to define standards and consolidate server platforms/types – we will always have application specific requirements, but for the commodity kit, can we not standardize models, configurations agree a set of standards, that Janet always buys ML370s and not DL380′s is admirable for Janet, but not for our operating costs, drivers, firmware and support etc.
• IT to establish workshops and meetings to discuss data center capacity issues – to involve the client in the standards get the buy in and support at the top level.
• To look at virtualization of the infrastructure and the application – is there a reason we have so many physical servers? How do we get buy in and investment to deliver value?
• To look at consolidation of the applications with a viewpoint of moving to shared infrastructure particularly on those quick wins, Citrix or web, reducing the number of individual departmental specific Citrix and Web servers which might not all be busy or nearing 50% capacity most of the time.

Key though to the discussion is onboarding IT (both the support and application/development teams) to the discussion, establish where we are, where we want to go and work on the basis that we need to be more efficient in the way we code, deploy and configure the infrastructure. Illustrate the performance per watt discussion,  will that extra 300MHz at 45 more watt deliver 300MHz or 45w worth of performance per processor?

Share and Enjoy

I’ve been speaking with Chris over the last few days to see how he is down in sunny Canary Wharf and I admit to see if his organization is recruiting, (I’m looking for work), anyway one of the things I was talking to him was about how IT recruitment and training is progressing. He was telling me the organization he works at was offering him more training on SMS or in Project Management, and I was asking him if we didn’t need to move training away from the technical arm towards more community, service orientated and interpersonal skills. He was not impressed, “less technical, but I’m a technical engineer”, yes but you see if I ask a Windows engineer (Chris) to look after some Linux servers (which is what happened to Chris at work), he went home, took a spare laptop or a virtual machine, downloaded RedHat and started using it. He might not have been enterprise specialized, or certified to the official standards, but within a week, he could manage services, create new disk partitions, create new user accounts and check the logs, the core activities he needed to perform.

But as we’re faced with cloud, with outsourcing, with an ever changing way of doing business, with more required from less, we need to manage expectations, manage our clients if you like, for this we need our technical teams to leave their technical desks, start communicating with the end user community. Enter a dialog if you like, be seen as permanently available, permanently improving your infrastructure, understanding your application, managing incidents, requests and changes to application and infrastructure like that task is the centre of their universe, to in essence add value. Be the difference between “..computer says no.. I can ping it, the server’s fine, (leave me to my MCSE book), because increasingly adding value whether it’s responding to calls within the SLA (playing the game if you will), not defending your corner or blaming other departments, but working as a unified IT team delivering service and adding value is where we should be going.

I’ll give you my favourite example to illustrate this: (the user was a senior business manager)

User phones IT and says his mouse is broken, the right click function doesn’t work, help desk log a call, engineer presses respond, leaves it for a few days without speaking to the user. User chases the call, helpdesk send an email to the team copying in the end user. Engineer ignores it because he’s busy. Helpdesk chase the team again, engineer turns up, visits the user looks at the mouse, tries it and says, “it appears your right mouse button is broken, we don’t have any computer mice in stock, (he hasn’t checked and in any case believes the end user should pay their true cost), you will need to raise a purchase order for a new HP USB scroll mouse.” User asks how long this will take, well we’ll close this incident and raise a request for a new mouse purchase call, that’ll be a few weeks (he then meets his SLA).

A new call is raised for the purchase of a new HP USB Scroll mouse with the specified part number for the Compaq Evo desktop. The call is passed to desktop to check compatibility, it waits a few days, an engineer turns up, looks at the pc and goes, well it’s got USB, that’ll work fine, asks the user if there is a particular mouse he requires, user says “just one with right click”, engineer fills in the form, passes the call to purchasing.  Purchasing send out a purchase quotation request for a computer mouse, requesting specifically this part number, this particular USB Scroll Mouse in that particular colour as per the request, and get a quote for £45. They then create a purchase approval form for them to order the mouse if the manager is happy to pay the cost.

It’s been 12 working days since the original incident . In the meantime, the user calls help desk, “I have a Logitech USB mouse which I bought at PC World, can I use that? I really want a working right mouse button”, help desk ask user to wait, call the engineer, “Sorry a Logitech mouse wont work, you can only use official devices with your pc, we have to use a HP mouse, besides Logitech mice are not supported under the support contract, we must use supported devices”. The following email is sent:

“Dear…

Your purchase request for a HP USB Scroll Mouse in carbon part number DC172B is with purchasing, call number 00148673. (The link contains the mouse specifications)

The Logitech mouse is not compatible with your HP PC.

Additionally, the mouse is not covered by our support contract and it is against IT policy to plug in external devices which are not validated by IT Security and IT Support.  We can submit your device for electrical testing, and then have it validated against the desktop build to check the drivers and software are compatible.  Please advise me how you would like to proceed.

Kind regards

Jane

Customer Delivery Specialist

Help Desk Team

Meanwhile, the users’ manager who runs the front office trading team, gets a purchase approval request looks at it, “£45 for a mouse? That’s expensive, why so much? I’m not paying that, why can’t IT just give me a mouse?” and rejects it. User is told electronically that his mouse has not been approved, gets upset and emails the universe saying that IT are rubbish. The CIO gets an email from the head of trading asking do we have a mouse, CIO calls head of desktop support and demands a mouse is taken to the user, asking “do we have any computer mice?”, “of course, about 35 of them, we just got new pcs delivered, I’ll take one down to the user right now”

In the case above, we were teaching our IT teams about cost centres, about users paying their true cost, about delivering through the process and meeting service level agreements, about ‘playing the game’, and in essence of following technical processes. The engineer technically was not wrong, IT standards stated that you could not plug in devices that weren’t authorised by IT security, and they need to be tested for health and safety, the existing computer mouse was beyond economical repair and not covered by warranty, technically and operationally a new mouse should be purchased. Technically installing a Dell mouse would require you to install Dell drivers or software which is not part of the desktop build, so the drivers would need tested and packaged.

However, had we changed the mindset from technical, from cost center and paying their true cost to one of delivery, reducing the support cost, it would have been apparent that:

• Two calls had been logged
• Two engineers had spent about an hour of their time
  • One looking at the original incident – right mouse doesn’t work
  • Another checking the pc model, checking the mouse part number, discussing the specifications with the user, did they want an optical mouse, did they want more than two buttons, did they want USB, their pc also supported PS2, also did the colour of the mouse matter, they come in carbon or black.
• Purchasing had created a purchase request, had got three independent quotes for a HP USB Scroll Mouse and created an online approval request for the mouse
• The CIO had been diverted from strategy, from delivering value to the business and got told, your priority is to get Bill a new HP USB Scoll mouse

We have in essence:

• Duplicated effort
• Created internal cost and workloads
• Damaged end user perception of delivery, of being obstructive rather than adding value

Share and Enjoy

PC World

Dell on Wednesday said it would offer pre-configured systems for enterprise customers looking to get server environments up and running quickly.

The fixed configurations include servers, storage modules and software that are pretested and can be deployed in hours, Dell officials said. The offerings are a diversification from Dell’s traditional built-to-order business model, in which customers typically specify configurations before ordering.

Dell is offering fixed systems as an option to custom-built systems that could take longer to configure and deploy, said Praveen Asthana, Dell’s director of enterprise storage.

“The goal is speed,” Asthana said. “Now we’re saying you can be up and running in 30 minutes, not in a week.”

The systems are targeted at customers like small- and- medium businesses who may lack expertise in server deployment, or to those who want to quickly deploy servers with applications like virtualization.

I can certainly see the appeal of this, having pre-defined configurations aimed with specific markets or uses can aid the sales process and improve Dell’s ability to ship out systems more quickly, anything we can do to aid the end user community in ordering the right technologies, in making the technology more affordable and accessible has to be a good thing.

Interestingly this was something that I was speaking with at an enterprise in the city, which had taken to consolidate the range of servers they bought, reducing the I’ll have one of these to a very simple, small/medium and large configuration, (anything else was then looked at on a per project basis).

Small might be:

• 2 processors
• 8GB RAM
• 4x300GB drives

Medium might be:

• 4 socket processors
• 16GB RAM
• 2x 300GB drives and SAN storage

Large might be:

• 4 socket or 8 socket processors
• 32-64GB RAM
• 2x 300gb drives and SAN storage

By doing this, the company was able to negotiate more standardized (stable) prices through their purchase arrangements, reduce the time it took to get a server shipped, and reduce the different types of x86 servers, from 16 to 3, significantly reducing their support cost, driver and firmware management activities, as well as knowledge that the engineers had to have per platform. More crucially it met the majority of end user requests, people understood the concept immediately, and it allowed IT to choose the configurations more easily without debate, making the process more binary, the debate moved on from drive capacity, server models to whether I needed two sockets or four, whether I needed SAN storage or not. Interestingly the same concept is being tried in their virtual configurations next month.

Share and Enjoy

Techtarget

Last week, the FC-BB-5 working group of the T11 Technical Committee unanimously approved a final standard for Fibre Channel over Ethernet (FCoE). Now the technology’s champion in the storage marketplace, Cisco Systems Inc., and some analysts expect the formally approved standard to drive adoption of the fledgling standard in the market.

Claudio DeSanti, a distinguished engineer at Cisco Systems and the T11 committee chairman, said there were no no significant changes to the specification for the standard between establishing “technical stability” with an earlier draft last fall and the final ratification. However, there were four members of the committee who voted No on the earlier version and then changed their votes to Yes on the final standard after “resolving communication” and clarifying some points about how the spec was written.

Even though vendors, including Cisco Systems, have already begun rolling out FCoE products based on the “technical stability” version of the standard, Tam Dell’Oro, founder and president at anaysts Dell’Oro Group, said a lack of an official standard has been known to stall adoption of new technologies in the networking market.

She cited the example of the IEEE 802.11 standard for wireless routers. Dell’Oro said her research has shown that an older, well-established version of the standard called 802.11g still accounts for 98% of the products purchased by telecom service providers over the 802.11n specification. “If enterprise users behave that way with that type of equipment, it would probably be the same with [FCoE],” she said.

An article talking about Fibre Channel over Ethernet, which has been a topic of debate with colleagues for a number of reasons. Unifying the storage and network down one pipe makes sense, it should further enable a reduction in the time it takes to deploy storage or network capacity in line with the business need, it should simplify connectivity by reducing the number and variants of patching and possibly empower or on-board new users to look at SAN storage, 10GB Ethernet etc, I might be buying kit which is enabled for this new medium. That said there remains the legacy infrastructure, the 100MB or even 10MB Ethernet networks which work, which deliver everything the end user needs and they wont be switched overnight, there are the enterprises that have invested in fibre big time and are unlikely to switch overnight, but let us not get too caught up in negative language or thoughts.

Is Fibre Channel over Ethernet for everyone? No but then neither is 10GB Ethernet, Grid, virtualization or solid state disk dependent on your technical and business requirements and constraints. What we will see though is that it will have every opportunity of becoming the default standard, that your server becomes Fibre Channel over Ethernet enabled, just like your server jumped to 1GB standard, as the cost of the adaptors falls, as more people realize the benefits of converging storage and network down the same connection for virtualization, for systems management and deployment, so it becomes available out of the box, but we’re some time away. What we need and we’re seeing that now, is the industry wide standards both technically, and in terms of best practice configurations/guides, that I know when I buy an adaptor for my server that it will work with any switch using agreed common standards, once we have that, coupled with multiple vendors, then we can get the econmies of scale, the comeptition in the market place to drive down the cost and drive up the adoption, exciting times are ahead.

Share and Enjoy

SGI

FREMONT, Calif.—June 15, 2009—SGI® (NASDAQ: SGI) will showcase several of its new energy-efficient data center technologies at the Gartner IT Infrastructure, Operations & Management Summit 2009, taking place from June 23rd to 25th in Orlando, FL, in Booth N. SGI will also present best practices in cluster computing, including a unique look into technology innovation at the cabinet level.

SGI Speaking Session: “Breakthrough Density Energy-Efficiency for Cluster Computing,” presented by Geoffrey Noer, senior director of product marketing at SGI.

Date: Thursday, June 25

Time: 11:00-11:30 a.m. EDT

Location: Osceola Ballroom 4-6, Gaylord Palms Resort and Convention Center

Topic: Geoffrey Noer will discuss CloudRackâ„¢ C2, a solution that delivers dramatic bottom-line savings by assuring maximum power usage and cooling efficiency, as well as offering staggering server densities with up to 1,280 cores per cabinet. CloudRack C2 capitalizes on an ultra-efficient, rack-centric Eco-Logicalâ„¢ design, which uses fanless and coverless 1U server trays installed in 24-inch wide 23U and 46U intelligent cabinet configurations.

It will be interesting to see/hear about these new servers this week from SGI, anything they can do to improve systems management and energy efficiency, whilst balancing performance has to be good for innovation of the server and choice for the end user community. I wonder what inovations we will see and if there is anything in terms of virtualization, I’ll need to check it out.

Share and Enjoy

PRNewswire

MINNEAPOLIS, June 15 /PRNewswire/ — NetEx(R), the leader in high-speed WAN optimization software, today announced an extension of its support for the VMware virtual environments by qualifying VMware vCenter Site Recovery Manager on its award-winning HyperIP(R) for VMware WAN optimization software.

The HyperIP-Site Recovery Manager solution automates and accelerates disaster recovery operations to minimize down time and ensure that enterprise users can meet their Recovery Time Objectives (RTO) and Recovery Point Objectives (RPO).

VMware vCenter Site Recovery Manager accelerates recovery processes by automating the execution of failover and simplifies the creation and management of recovery plans. The DR plans become an integrated part of the VMware virtual environment. In the event of a failure, Site Recovery Manager automatically executes the recovery plan while administrators retain full control with the ability to pause or stop the execution at any point.

HyperIP for VMware is a software-only implementation of NetEx’s award-winning WAN optimization software for disaster recovery optimization of backup, recovery and data replication applications, enabling unmatched flexibility in deployment and performance scalability on virtual machines. HyperIP delivers unmatched performance in accelerating DR operations by moving TCP data more efficiently using patent-pending technology. It accelerates and optimizes industry-leading VMware-enabled data replication and file transfer applications – including VMware VMotion, Veeam replicator, IBM TSM, Data Domain replication, DataCore AIM, and others – by aggregating multiple data replication applications over a shared connection while minimizing the effects of network latency and network disruption. This makes HyperIP for VMware an ideal complement to vCenter Site Recovery Manager for optimizing enterprise DR processes.

Anything the vendors can do to further enable high availability and business continuity in the virtual world has to be a good thing, and it’s something I’ll need to check out.

Share and Enjoy

I got a fact sheet from Cisco talking about Fibre Channel over Ethernet which has been in the news recently relating to new standards being submitted for approval. The fact sheet was very cool, it highlights some interesting benefit and cost illustrations over a traditional setup. The cost improvements and energy efficiency savings are impressive according to the document. As with any statement about costs and efficiency they can be debated, so if we abstract ourselves from values costs and talk about converging storage and network through the same connection, we can establish whether it’s going to be the next big thing or if it’s still some time for the future.

We’ve seen a lot in the last few years with 10GB Ethernet, originally I remember colleagues saying “you’ll never need 10GB…”, only to find them talking about it being very useful as the backbone platform for their virtualization or hpc solutions, and at the moment I think Fibre Channel over Ethernet is at a similar stage. The current comments being about “we’ve invested in fibre…”, or “we’re not upgrading our switches or converging storage and networks, it’s too expensive…”.

That’s right, at the moment their existing network and SAN infrastructure is in place, the patching is there, everyone understands their place in the infrastructure world. But long term, Ethernet is fairly understood as a medium for communication in the workplace, taking it therefore to the ‘next generation’ of supplying the storage and the network down the same converged pipe isn’t necessarily world ending, and could easily be an extension of the platform like 1GB over the 100MB which has been standard for years, and still is in many an enterprise.

Let us not get negative on an idea, on a concept of new technology, let us see what it can bring, let more companies onboard new products, solutions and innovations around Fibre Channel over Ethernet, continue working on standards, drive down the costs, improve our ability to manage the infrastructure and realize Fibre Channel over Ethernet as a business or communication enabler, rather than that difficult word or concept, change.

In essence, we’re de-duplicating workloads, roles you might say but not people.  The current process for a server build might be:

1. Allocate server host name
2. Allocate IP address and network port
3. Patching supply patch cable
4. Allocate storage ports
5. Data patching supply patch cables
6. Operating system team install operating system
7. Storage to zone in the cards to storage

If we can change this to: (where networks can allocate ports, and the patching team can add another cable(s) where necessary)

1. Allocate server host name
2. Allocate IP address, network ports and storage ports
3. Patching supply patch cables
4. Operating system team install the operating system
5. Storage present the storage

Remember from an end user point of view, the fewer teams involved, the fewer steps on the call to get my server installed, the less time it takes for the server to be delivered.

Share and Enjoy