keeping simple

future of the data center

This article from Ars Technica discusses a talk over the summer by Merrill Lynch’s chief technology architect, Jeffrey Birnbaum on “stateless cloud computing” - most concretely on distributed file systems.

Birnbaum believes that one of the key foundational elements of a stateless computing environment is a networked storage system that enables ubiquitous availability of software. The file paths of the individual applications should be based on clearly defined nomenclature, much like the domain of a web site. All application dependencies should be accessible through the network filesystem, and version numbers should be expressed with the path nomenclature.

Big distributed file system - sure. Why should version numbers be expressed with the path nomenclature (a Plan9 idea, btw)?  Now we go on to the ancient problem of caching distributed data.

The obvious challenge posed by rolling out worldwide network storage infrastructure is scalability. If everyone in a global organization is depending on a network storage solution, then it needs to be fast and consistently reliable. The solution that Birnbaum proposes is regional mirroring and caching. The storage system would be universally synchronized between mirrors that have all the data. Caching can also be used at individual facilities to further improve performance. To achieve this kind of global scalability, he says, the best approach is similar to that of Akamai.

So even with a non-globally distributed file system, the problem of shared access is non-trivial. A global file system makes things quite challenging. Suppose we have a file recording trades and the Singapore, London, NY, and Espanola main offices all are reading and writing at the same time. Caching and cache coherency is an utter nightmare.  Akamai, like Google, solves the problem of massive amounts of distributed data by focusing on “delivery” - otherwise known as “read only content” or “many readers one writer” and with no requirement for true synchronization.  But the ML problem is more difficult even if we ignore multiple writers because, presumably, you want Singapore to actually see every trade made in Espanola even though for Akamai, it’s ok if the cache is not fresh. How to solve multiple readers and writers is something else as well.

These concepts don’t cover a whole lot of new ground yet. Much of this was already possible with conventional thin-client systems. The point at which it becomes immensely valuable, according to Birnbaum, is when all of these technologies are used together with virtualization to abstract the processes away from the hardware. Once this is done, individual operations can seamlessly float around data centers and balance out in a manner that offers a more optimal level of resource utilization.

And this seems to me to gloss over the even harder problem. Imagine a serious Oracle application “seamlessly floating” from some set of machines in one data-center to another set.  I can’t imagine how that works. Imagining little jobs floating is easier, but is that really an interesting problem? And this brings us to the most interesting claim:

He claims that 61 percent of a company’s enterprise server capacity goes completely unused and proposes an automated load balancing solution—

SIXTY ONE PERCENT!  Think of the power use.

Operating system research - 16 years perspective

It’s somewhat funny and somewhat sad to read this thread on the old USENET. Starting out with Andy Tanenbaum’s proposed list of accepted truths (most of which I thought wrong at the time)

GENERALLY ACCEPTED AS TRUE BY RESEARCHERS IN DISTRIBUTED SYSTEMS

• - The client-server paradigm is a good one
• - Microkernels are the way to go
• - UNIX can be successfully run as an application program
• - RPC is a good idea to base your system on
• - Atomic group communication (broadcast) is highly useful
• - Caching at the file server is definitely worth doing
• - File server replication is an idea whose time has come
• - Message passing is too primitive for application programmers to use
• - Synchronous (blocking) communication is easier to use than asynchronous
• - New languages are needed for writing distributed/parallel applications
• - Distributed shared memory in one form or another is a convenient model

and then Rob Pike’s refutation entitled “Andy Tanenbaum hasn’t learned anything”. A key point that comes up later in the discussion is Tanenbaum’s (incorrect) assertion that a “factor of two” performace loss is nothing to worry about. The date of the discussion is interesting, because in a few years the Linux Tsunami washed away most of the landscape of this discussion. As for my contribution it is very disturbing to see that I have not learned much about those topics in the last 14 16 years.

It is clear that Rob Pike was right on many more issues than AST, but that seemed clear at the time too.

[edited to reflect the passage of time]

Doing business with Avaya’s bizzare business model

Years ago, we got some equipment from Avaya - not terrible, but not wonderful. A junior tech with no signature authority signed a contract with them that says in big letters on the top “until 2005″. In 2007 we got rid of the equipment and now have spent over a year attempting to cancel the contract - getting one absurd response after another from Avaya’s abusive collections department.  I’d never do business with these people ever again and I advise anyone else considering using their equipment to get all contracts carefully reviewed. What an attitude!

Reliable broadcast algorithms: after 20 years

New paper with short description of the Chang/Maxmchuk algorithm I have been championing for 20 years or more.