The next generation of high-rate server connectivity and core switching is with us. IEEE 802.3ba, a new standard governing 40 Gb/s and 100 Gb/s Ethernet operations, and the first to cover two specifications simultaneously, has been ratified. According to an IEEE statement:

“IEEE 802.3ba will eliminate these bottlenecks by providing a robust, scalable architecture for meeting current bandwidth requirements and laying a solid foundation for future Ethernet speed increases.”

The new standard:

“addresses critical challenges facing technology providers today, such as the growing number of applications with demonstrated bandwidth needs far exceeding existing Ethernet capabilities, by providing a larger, more durable bandwidth pipeline. Furthermore, collaboration between the IEEE P802.3ba 40Gb/s and 100Gb/s Ethernet Task Force and the International Telecommunication Union’s Telecommunication Standardization Sector (ITU-T) Study Group 15 ensures these new Ethernet rates are transportable over optical transport networks.

John D’Ambrosia, Chair, IEEE P802.3ba Task Force, and Director, Ethernet-based Standards, CTO Office, Force10 Networks is quoted as saying:

“Ubiquitous adoption of bandwidth-intensive technologies and applications, such as converged network services, video-on-demand, and social networking, is producing rapidly increasing demand for higher-rate throughput. As mass-market access to these technologies continues accelerating, coupled with today’s progressively more powerful server architectures, data centers, network providers and end-users alike are finding themselves confronted by pressing bandwidth bottlenecks, IEEE 802.3ba will eliminate these bottlenecks by providing a robust, scalable architecture for meeting current bandwidth requirements and laying a solid foundation for future Ethernet speed increases.”

The arrival of 802.3ba will drive new network and service developments and the aggregation of 10Gig network deployments, and will be a shot in the arm for the delivery of faster broadband access. In the words of David Law, Chair, IEEE 802.3 Working Group

“This is truly a forward-looking standard that will spur innovation at every point along the Ethernet value chain, as well as providing the essential architecture needed to facilitate greater broadband connectivity on a global scale, IEEE 802.3ba ensures that we can meet today’s needs while preparing for the next generation of emerging technology developments.”

802.3ba is backwards-compatible with existing 802.3 installations, preserving existing investments in multi-Gigabit infrastructures. The new technology will also lower operating costs and increase energy efficiency by simplifying complex link integration schema used in today’s network architectures.

A little bit of fun for non-footie fans and those already bored of watching men in colourful shirts and gaudy footwear chasing a beach ball around a field.

This quiz is brought to you by those nice people at How Stuff Works?.

ADD YOUR SCORE TO THE COMMENTS BOX.

JUST A BIT OF FUN THIS TIME. NEXT TIME, WE’LL BE RUNNING OUR OWN COMPETITION, AND THERE’LL BE A TASTY PRIZE UP FOR GRABS

by Steve Barnett of VMC, and originally published on their VM Yak blog:

Hardware today allows for higher and higher consolidation ratios “even if it is not our hardware”. More and more dense ram configurations are available but according to IDC the average number of VMs run on a single piece of physical hardware remains stubbornly at ’seven’.

Economics means we’re incentivised to increase this ratio but fear stops us from doing so.

The concentration of risk that comes from the consolidation of more and more applications/ workloads/ services, call them what you will, onto fewer and fewer machines runs completely contrary to everything we have learnt to do in the last 20 years.

But, to our predecessors in the 80’s and 90’s this was considered normal everything ran on the Mainframe, a couple of SuperMini’s or on half a dozen VAXs.

What’s happened is that we have been fooled into believing that the answer to needing to run a new application or service is to buy a new cheap server, and lets face it it’s never been cheaper to buy servers – it’s the spiraling cost of running them that is killing us!!

Do you think that the major Cloud providers are using 10’s of thousands of those hot little 2CPU 16GB systems that DELL & HP are shoving down everyone’s throats?

No, they are built on RAM heavy, massively scalable ‘custom designed hardware’ with ‘custom designed’ capacity management software. With, of course, ‘custom designed support’.

Just what we’re offering to you, everyday… only with the Virtual Machine Company it’s our standard design..!

Steve at Virtual Machine Co. has just sent us a link to this item on theregister.com about IDC’s report on server sales figures for Q1 2010, and provides his take on what they reveal.

First-quarter 2010 server sales figures released by IDC show factory gate prices up 4.6% to $10.42 billion, meaning server sales in 2010 will probably exceed $40 billion, a return to the revenues manufacturers enjoyed in 2007.

X86-64 servers running Windows Server are driving the growth, being installed on 47% of all servers worldwide. In particular, Windows Server is present on larger systems, and this is driven by the new datacentre licensing model that Microsoft has adopted which allows specifically for large-scale virtualisation.

Linux sales have grown substantially ahead of the market, while more traditional UNIX systems have seen a contraction with non-x86-64 based systems significantly losing market share.

Falling sales of systems based on Itanium and RISC chips will be particularly hurting IBM on unit sales, but IBM continues to grow their services business. And where IBM go the rest will follow. You only have to think about the acquisition of EDS by HP, PEROT Systems by Dell and Sun Field Services by Oracle to see that the future of all of these previously hardware focused companies is in the direct sale of products & services to major enterprises, governments etc., leaving the mid-market and smaller enterprises to be supported by the channel. As for SMEs, they will have to make do with buying off the web and receiving support from user forums in the same way.

Gartner Group published their Q1 server sales figures last week, and whilst they include channel sales (so these figures will include stock in the channel), they are pretty much in line with IDC, showing overall recovery in platform sales. For more on the Gartner report, follow this link to another article on theregister.com. You can Google around and find the tabulated set of figures.

But the big news is that in the first quarter of this year, 2 million servers shipped worldwide. If we’re to believe the figures previously published by IDC nearly half of those will be virtualised, and with an average consolidation ratio of 6:1 that means that in the first quarter of this year 6 million new virtual machines came online.

As covered on theregister.com this morning, J P Morgan analyst Mark Moscowitz has published a report showing that NetApp is creating clear water between itself and its nearest rivals in the virtualisation market thanks to better integration with Cisco, VMware & Microsoft.

It’s a trend that’s set to grow, though probably not to such an extent that NetApp achieves dominance. However, the report also indicates that NetApp is second only to Apple as a share growth prospect.

EACS Media has been working with id7 Media Solutions on a small footprint, location-based ingest station. It’s a self-contained, transportable touch-screen device, accessible via Ethernet – CIFS or NFS – from a Mac or a Windows PC. There are several variants in the range, with useable capacities of 2TB to 10TB and the option of an on-board LTO 4 or LTO 5 drive.

Files can be copies from location media to disk to tape simultaneously. Tape-to-tape duplication is also supported.

Get in touch if you’d like to know more. We’ll have demo units available at the end of May if you want to test one in anger!

Although you probably wouldn’t consider using the Amazon Simple Storage Service (S3) ‘public cloud’ to back up your professional media projects, Amazon’s explanation of how the service works, and even the pricing to some extent, is a decent guide to how clouds can fit into a business’s storage, backup and archiving strategy. Here’s an extract from the introduction to the service:

Amazon S3 provides a simple web services interface that can be used to store and retrieve any amount of data, at any time, from anywhere on the web. It gives any developer access to the same highly scalable, reliable, secure, fast, inexpensive infrastructure that Amazon uses to run its own global network of web sites.

Storage capacity on S3 is unlimited, and each object stored can occupy up to 5GB of space. Objects are stored in ‘buckets’ which can be located in one of a number of regions – US Standard, EU (Ireland), US West (Northern California) and Asia Pacific (Singapore) – for easy access with minimum latency using a unique, self-generated key. Objects never leave the region where you choose to store them unless you choose to move them.

Iron Mountain also offers a similar Storage-as-a-Service (StaaS), in their case to businesses.

If you were to consider storing your valuable media data off-site, what would you expect in terms of SLAs, security, access, flexibility etc. from your service provider?

Despite opinions that they’re destined to fail, Toshiba is considering producing hybrid 2.5-inch solid state / hard disk drives, reports The Register. In the opinion of Zsolt Kerekes of StorageSearch:

“the volume market for these would be in enterprise arrays – but you get much better results by mixing and matching genuine SSDs and genuine HDDs with an ASAP SSD type of controller (or human tuning) and at much lower cost”.

An ASAP controller, such as Dataram’s XcelaSAN, according the Kerekes:

“… enables users to get usable speedups from (separate) SSD caches running in tandem with HDD arrays in minutes or hours without human hot spot tuning”.

The problem area is identifying data hot spots quickly enough and moving them from the HDD component to the SSD component of the drive. ASAP controllers do this automatically.

Kerekes believes Flash drives with a massive RAM cache are viable as hybrids, and do meet real needs for application acceleration.

The interconnects between nodes in High Performance Computing (HPC) environments has always posed one of the greatest challenges when putting together a fast, reliable system. Node-to-node bandwidth, and bandwidth to all nodes at the same time requires high speed switches and complex topologies. Equally important are latency between nodes, protocol handling, application support, in-built intelligence and, last of all, affordability.

Infiniband has met most, but not all of these requirements, as have some less well-known proprietary interconnects. But it’s 10 Gigabit Ethernet that’s set to steal the show, as this article on theinquirer.net explains.

As revealed in this item on the datacenterknowledge.com website, the volume of digital information created and duplicated around the world will hit a staggering 1.2 zettabytes in 2010 (that’s 1.2 million petabytes), exceeding forecasts, and is set to grow to 35 zettabytes by 2020. This is according to research conducted by IDC.

Over 75% of what we store is a copy. The number of files being managed will grow by a factor of 67, and the staffing required to keep up will grow by a factor of 1.4. It’s no surprise the report makes a strong case for moving towards a Digital Universe that is extensively de-duplicated, and sees automation and management tools as key to tackling this challenge, along with a gradual migration to cloud platforms that offer scalability and favorable economics.

The Digital Universe will create 1.2 zettabytes of data in 2010, and this is set to grow to 35 zettabytes by 2020.