Taylor's Take on Sun Storage
My storage team and I focus on three of the most important aspects in any industry: customers, competitors and market trends. There is insight to gain and share in this role, so here is our take on Sun and Storage - Taylor Allis
Open Storage: Early Markets
As promised, below is a second White Paper on Open Storage. This second one addresses Open Storage market drivers and growth and is titled Open Storage Adoption. It can be downloaded here:
Again, I would like to thank Bruce Norikane, our Sr. Analyst, for
his help as well as our market research manager, Chris Ilg, for his forecasting work. And again, I'll use this blog to post the CliffsNotes for those short on time. Below I will cover the need for a new storage architectures and early target markets. In subsequent blogs I'll cover the Open Storage future market forecast, other vendor initiatives and customer case studies - early adopters who have used Open Storage to solve their critical business needs...
The Need for a New Storage Architecture
Bruce, mentioned above, made a profound statement during our Open Storage planning that ended up in the White Paper. He said, "Google and Amazon would not exist if they hadn’t built their own storage infrastructures." They certainly wouldn't exist in their current state. When they started, traditional storage architectures were too expensive and inflexible to support the business model they had in mind. So what did they do? They had to buy commodity components and developed their own software like the Google File System (GFS).
Certainly not everyone can build their own file system today. But the requirements that drove Google to build their own file system have done nothing but increased. Consider the following facts:
- IDC’s report “The Expanding Digital Universe" states that new digital data will grow 6X between 2006 and 2010 (Emerging market data is growing up to 40% faster)
- 70% of this data will be created by individuals (me, on this blog for example)
- Enterprises will be responsible for storing 85% of this new digital data (Sun's IT dept has to manage the data I create)
- YouTube was founded in February 2005, and by May 2006, 50,000 videos were being added every day
- A YouTube search today can return about 72.1 million videos and 2.89 million user channels
A new, more economic and scalable storage architecture is desperately needed - enter Open Storage...
Open Storage Growth Markets
Open Storage can (and will) compete with traditional storage architectures. But Open Storage won't "take over the world" overnight. Most likely the data center mix of open storage architectures vs. closed storage architectures will change over time and vary data center to data center (if history is our guide). But what markets will adopt sooner? What are the Open Storage "sweet spots"?
Web 2.0: I count Web 2.0 apps as applications delivered via the Web. Apps like blogs, wikis, podcasts, RSS feeds, mashups, and social-networking sites like MySpace, Facebook or SmugMug. Consider this:
- Technorati stated in March 2008 that there were 112.8 million blogs
- Bloggers update their content with 1.6 million posts EVERY day (or more than 18 updates per second!)
- There are currently 3,115 mashup applications on the Internet
And Web 2.0 apps are not just for up starts - Forrester surveyed 2,200 IT decision makers from traditional enterprises and found that 33% were planning on investing in Web 2.0 applications. Web 2.0 storage requirements differ from traditional storage requirements as well. They need massively scalable but low-cost systems. Web 2.0 users are even willing to trade high availability for lower costs. Everyone needs high scalability at lower costs - but the need in the Web 2.0 space is acute. Thus, Web 2.0 will be the key driver for Open Storage architectures.
HPC Storage: IDC estimates that HPC storage systems added about $3.9 billion to the 2006 server revenue total and will undergo faster annual growth than HPC servers. Maximizing I/O bandwidth and minimizing latency while scaling storage capacity is the top priority for HPC storage users. Because of this, data locality is an issue for many HPC implementations. What's data locality? HPC services provider Instrumental, Inc.explains:
Data locality is a big issue in some architectures. Sometimes you need to know where data is in memory to get the best performance. Locality issues are compounded by the enormous amount of software ‘in the middle (OS, file system, volume management, failover, host bus adapters, and so on)’.
To manage issues such as data locality, an open storage architecture is needed. The one thing that HPC storage deployments have in common is that they are all custom built. HPC users need direct access to their storage components and software along with the flexibility to swap components and customize software to optimize their storage. This is difficult to do with closed storage systems.
Additionally, parallel, shared or clustered file systems that leverage global namespace technologies are used in most HPC storage environments. This includes the HPC open source file systems Sun offers - like Lustre. In fact, Lustre is used in 15% of the top 500 supercomputers in the world and in six of the top 10 supercomputers.
Lastly, an additional top storage requirement in HPC is Hierarchal Storage Management (HSM) software (moving data from disk to tape). 
Why? Just look at the massive amounts of data HPC applications generate. The San Diego Supercomputer Center states their earthquake simulations alone generate 47TB every week! By 2011, they expect archived data to grow to more than 100PB. HPC centers must leverage the economics of tape to store such massive amounts of data. Sun offers tape as well as open-source HSM software for disk-to-tape data migration - Sun's Storage Archive Manager (SAM) software.
To see the real-world benefits an open storage architecture can offer HPC customers, see the Texas Advanced Computing Center (TACC) implementation of Sun Constellation - aka Ranger.
Server Virtualization: Open storage introduces more flexibility and consolidation benefits to the server-virtualization market. This added functionality can be realized in two ways:
- By running open storage software inside a virtual machine (VM)
- By running any vendor’s storage software on an open storage server
Storage users can now consolidate three servers and a storage appliance onto a single server. In a closed architecture, storage software cannot be separated from the hardware.
In the second scenario, users can use an open storage server, such as the Sun Fire X4500, as a storage target or shared appliance. What’s unique is that users can repurpose their storage appliance as their needs change. For example, customers can repurpose the same Sun Fire X4500 into a NAS device, a Virtual Tape Library (VTL) or a data replication appliance without buying more hardware. Now that's investment protection!
In the following diagram, a customer has taken a Sun Fire X4500 server running Linux-based VTL software and has repurposed it into a remote replication appliance by leveraging server virtualization and open source Sun StorageTek Availability Suite software.
Sever Virtualization and Open Storage can deliver better investment protection and significant cost and consolidation advantages ...
Next Blog...
What we predict the size and growth of the Open Storage will be
Posted at 04:06PM Jun 10, 2008 by Taylor Allis in Storage Intelligence | Comments[1]
Today's Page Hits: 36