enter the grid?

now it’s all about grid computing. according to oracle to be more precise. what a nice coincidence that they offer a new suite of products which enables you to just enter the grid ;)

if however you believe having heard of “grid computing” long long before, this probably wasn’t a mental delusion..

[update 20031002: there are pictures of the event online now]

an excerpt of my mental notes from the keynotes (my comments in [square brackets]):

– key drivers of future it value: increase productivity, decrease costs

– steps required: consolidation, standardization, automation

– gc facilitates small incremental growing using small and cheap, standardized, x86 based blades instead of large mainframes. start small and grow (or don’t).

– gc abstracts from and virtualizes operating systems and hardware lying underneath it all

– better reliability, no single point of failure [uh.. what about the grid abstraction software?]

– strong emphasis on cost saving by better utilizing existing hardware (instead of buying new one)

– oracle strongly counts on linux as linux is the first os to make cheap computing (required for grid computing) possible.

– grids will be the next “disruptive technology” (as was the www). this is meant positively in that it will make us think differently about computing. grids are key to managing immense data streams and volumes as seen at the cern

– according to intel, the gap between demand and available resources will get wider. new hardware [of course from intel ;)] is needed to catch up with the growing demand. intel promotes open architecture [hm.. and what about the long awaited and yet still missing publication of the specs of the wlan chip used for centrino?]

– the oracle 10g suite is made for grid computing (hence the “g”) while offering better performance than oracle 9i (“i” for “internet) and competitors [you wouldn’t believe this was an oracle event ;)].

..and so on. some noteworthy observations (my own, hehe ;):

– oracle really really bets on linux. windows was almost only mentioned in jokes. the same goes for x86 (including itanium2) as opposed to what they called “risc” computers (actually you can’t separate risc and cisc that easily anymore as most modern cpus use a combination of both technologies. however, “risc” was clearly targetted towards sparcs and ibm’s ‘power xy’ cpu line.)

– the accentuation of the (if true, quite impressive) performance boost between oracle 9i and 10g reminded me of the hard times ms had trying to sell windows xp to customers running windows 2k. the more you accentuate how much better the new version allegedly is, the more sceptical the people get about it (well, why would you need to advertise it that much if it really was that much better? ;).

– in recent years, security was one of the top topics with oracle (remember the “unbreakable linux” bet). now, security seems to be very much just a diminishing thing in the background again. this is astonishing as i’d expect security to be an even more important topic with grid computing than with single server systems. either they are (strangely) very confident that they fixed it once and for all (impossible by nature for such a complex system) or they just jump on everything which looks like a nice marketing hype occasion (are average oracle customers that hype-driven really?).

– the emphasis on “start small” and the focus on decreasing costs shows that oracle is willing to enlarge its potential market downwards (whereas for example ms with ms sql makes advances in the opposite direction). grid computing enables scalability, making oracle credible in a broader market.

– strong focus on the costs side. while reducing costs is usually desired, i do miss the perspectives on the “enabler/innovation” side. imho, cost reduction (or rather: efficiency increase) is inevitable but rather uninspiring in general (i hereby admit that i have a preference for creative minds ;). further, i believe extending business (besides reducing costs) will soon be the main focus of the people again.

– i personally would prefer distributed/grid computing at a lower (probably os-) level over just replacing a conventional dbms or application server by a “grid-enabled” one. the latter tends to look a bit like a “hack-around” to me. of course, we don’t have any major distributed os available atm (and oracle doesn’t sell oses anyway ;). pragmatic.

as a final note, the best received speech of the day (concluding from the applause by the audience) was the sound and very self-ironic one held by jamie shiers. [update: jamie’s slides (powerpoint) are online now]

overall a nice event. not as glamorous as expected (i would have added some more salt & pepper – read: surprising, funny, cool, unconventional, amazing moments) and a bit lengthy (almost 4 hours in total), but ok for these (rainy) days.


Leave a Reply

Your email address will not be published. Required fields are marked *

6 × 1 =

This site uses Akismet to reduce spam. Learn how your comment data is processed.