Tuesday Mar 10, 2009

Squaring the circle, from disruption to trust

Mike Shapiro is an expert in disruptive technology; he was working on Solaris in the early 2000s. He spoke to a number of us at Sun's Guillemont Park Campus about Amber Road, Sun's new disruptive file server technology. Sun and our customers have the opportunity to take advantage of the next big thing in network storage.

Mike explained that for a technology to be truly disruptive

  • it must be cheaper
  • it must be good enough
  • there must be a compelling reason for adoption

and Amber Road has two killer apps. Flash and Analytics. There is a small layer of functionality that Amber Road can't provide but the bulk of the market doesn't need it, and certainly doesn't need it at the price charged. Since storage is a trust business, Sun's storage sales teams and the customers need to understand very carefully the storage requirements. It is unlikely that any functionality not available is a universal requirement but in some cases, its not the right time for customers to move from their incumbent suppliers; they need some of the missing functionality. Talking to storage users about Sun's new storage concentrates the minds of everyone involved.

Over the last 10 years, there have been only two ways that the laws of physics and economics permit to make disk arrays faster, either increase the cache size, or increase the disk speed. The cost of Flash has dropped over the last three years, thanks to those of us buying mp3 players and pdas. The Amber Road box's software allow newly economic flash to do either or both. Sun is a leader in flash and certify enterprise flash for 3-5 years. and has additional advantages including the superior reliability of ZFS and the opensource pricing of the Unified Storage arrays. We don't licence a right to use. What we charge is based on what we ship, you don't get charged more as you turn on functionality. (This has nearly always been true of Sun, I remember when buying SunOS systems that one of the advantages was that network funtionality was bundled with UNIX where as I was asked to pay extra for networking and RAID functionality by my then incumbent supplier). Crucially Sun doesn't seek to tax its customer's innovation. The "no more to pay" approach also applies to the Analytics which come with the box and you can use them all. The software is available on a try before buy basis at www.sun.com and I will be downloading have downloaded it onto my laptop, see also Installing the Amberroad simulator above, to demonstrate to anyone that wants to see it. [ Sun's 7000 series storage simulator home page ].

Some of what is argued to be missing is FCAL support. Mike stated that the long-term winning strategy is to have only one cable going into the box. If there's to be only one winner, it ain't going to be FCAL; it needs to support Ethernet, and there's a demand for infiniband. Our proposed iscsi functionality release plans means that the Unified Storage boxes can offer block devices over the network and support for most enterprise data centres will only get better. Having said that, we propose to release FC target functionality in Q4 this year.

The value proposition for Amber Road is that its cheaper, good enough and offers game changing superior management. This often gets lost in a feature benefit analysis, which often seek to disguise what the features cost. Sun knows storage and can meet the trust requirements that customer's require, Amber Road shows that a trusted source can disrupt the economics, and its only the customers that win.

This was uploaded on 28th May 2009 and back dated to the date of occurrence, 11th March


Tuesday Oct 02, 2007

Slow cruise to Nevada

Despite having a very hectic week, I finally finished a Live Update on my Solaris partition and have enscripted it, so it shouldn't be so hard next time. I had to fix my zfs checkpointing routines as while I had turned it on to protect my work against stupidity, I wasn't deleting them. I have amended the Gnome startup scripts to invoke $HOME/.gnome.login and the .gnome.logout is also invoked appropriately, and .gnome.logout creates a snapshot. The main reason for doing this is that I am unsure that Gnome runs the login shells exit scripts, and not all shells have these. However my poor code also means that root won't start a JDS session any more. I am going to claim this as a security enhancement. :(

My interest and use of  ZFS and Live Update has all been inspired by Chris Gerhard who documents his ideas at The dot in ...---... Live Update is also a coming technology, so I think it's important for me to get to grips with it. 

I am now considering how to write a tcl/tk wrapper for my LU script as I am still a number of versions behind.This meant I still had to amend the grub menu.lst by hand and refresh the boot archive manually. (The latter requirement may have been caused by me, but now I know how easy it is to mount the shadow BE, I shan't leave it for so long again.) I must remember that the reboot command is probably best issued from a root failsafe session.

Tuesday Feb 28, 2006

An epiphany about ZFS

The hightlight of yesterday's conference to me was a presentation about ZFS. How long am I going to hang out for a british pronounciation Mysterious Smiley. The preso was delivered by Dave Brittle, Lori Alt & Tabriz Leman.

While much of the material delivered yesterday was standard "Dog & Pony" material, this version stayed away from the administrative management interface and while mentioning the ideological substitution of pool for volume, it concentrated on the transactional nature of the filesystem update, the versioning this enables and also "bringing the ZFS goodness to slash".

Somehow I suddenly get it. ZFS revolutionises the storage of disk data blocks and their meta data. It writes new blocks before deleting old one and so can roll back if the write errors. This also allows versioining to occur, the old superblock becomes a snapshot master superblock. The placement of parity data in the meta blocks (as opposed to creating additional leaf node blocks) means that error correction is safer and and offers richer functionally. More....

It seems to me that this technology will enable a sedimentation process to occur and that much of a DBMS's functionality can migrate to the operating system (or in this case file system). When I say much, when I first started working with DBMS (i.e. in the last century Smug Smiley), they often used the filesystem and often didn't use write ahead logs. By bringing this DBMS functionality to the file system, a process started by the adoption of direct & async i/o, the ZFS designers have closed a loop and borrowed from the DBMS designer's learning curve. Only the DBMS can "know" if two blocks are part of the same "success unit", but ZFS can implement a sucess unit and should begin to weaken the need for a write ahead log. It will also enable the safe(r) use of open source databases.

The versioning feature of the file system, when certified for use as a root file system will enable much safer and faster patching; it will enable snapshot and rollback . If system managers use these features to adopt a faster software technology refresh, then innovation will come to the data centre faster since newer code is better quality and should contain new usefull features. Disk cloning, snapshot and rollback will also enable the rapid spawing of Solaris Containers. Fantastic.

We are also released from the tyranny of the partition table, which for the last 15 years we have required a volume manager for.

Despite these fantastic advances, when it becomes available, it'll be a V1.0 product, so care will be needed. Certainly, the authors seem to have some humility about this, but with Solaris Express, we can get hold of it now and begin acceptance and confidence testing. A final really great feature is that ZFS has been donated/incorporated into OpenSolaris.

This stuff should be available as an update in Solaris 10, maybe sometime over the summer and I'm going to get hold of an "Express" version for my laptop.

Edited A correspondent called Igor asked for a link to the slides. OpenSolaris has a documentation page which hosts a .pdf presentation.





« July 2016