CloudSlam Day 2
I’m ready to liveblog the second day’s events at CloudSlam’09. The first session starts at 8am Eastern, so here I am, dutifully logged in to WebEx and plugged in to the conference call, at 4:59am Pacific. And I haven’t even had any coffee…..
The first session is by William Fellows of the 451 Group, entitled “The Sky’s The Limit”. Industry analysis: part of 451′s ICE practice. Interesting nuggets: while eco-efficiency has been displaced by “CapEx to OpEx” as the main driver, he expects that “green IT” legislation will be coming. Nice example of Bechtel’s “internal cloud” (even though they don’t use the term): 30% infrastructure cost saving.
Here’s an interesting thought: “‘cloud’ enables execs to talk about big concepts in IT”. CEOs and CIOs will understand just enough to be dangerous….
7:00am Back at my desk, fortified by a quad espresso macchiato from Starbucks, it’s time for David Bernstein’s talk on “Cloud Building: Secrets from the Vault”. I have a visceral reaction against books, talks, articles, etc. where the title includes the word “secrets” – if it’s so damned secret, how come you’re publishing it? – but I’ll suspend judgement… David is Cisco’s “VP of Cloud Computing”. [...] Yes, the complexity bottleneck is the traditional (1GbE) network mess, and the lack of virtualization at the networking layer. So David’s pitching tag-based VN-Link over unified 10GbE, supporting VMotion in a way that’s consistent at the network and hypervisor levels. [...] Impressive pitch for UCS. Sample config goes from 75 switches to just 4. But this talk was advertised as revealing Cisco experience in working with cloud vendors to create existing clouds. VN-Link and UCS are future, right?
(I asked about load balancing in the cloud, which led to an interesting set of points.)
8:00am: Winston Bumpus of VMWare and the DMTF on cloud standards. Routine “what is a cloud” preamble; good breadth-first survey of the relevant standards activity; deep dive on OVF. (Not surprising, given VMWare’s involvement in the work.) In response to my question about whether standardization is premature, Winston’s position is that it’s “never too early to standardize. Never? Hmmm.
9:00am: Now one of the sessions I’ve been eagerly anticipating: Janine Anthony Bowen and Steve Sorrett on “Legal Issues in Cloud Computing”. When your data is commingled on a physical device with that of other parties, things can get tricky. I hope they address this. [...] Their agenda is: Privacy and Security, Legal and Business Contracts, and Government Contracting. This stuff is essential: anyone involved in hosted services of any kind needs to understand it. Outstanding.
10:00am: Keynote from Russ Daniels, Cloud CTO of HP. OK, he’s avoided bringing the cloud down to earth (turning it into fog), but he’s fallen into the opposite trap of going so high-level that it’s just “the future of the Internet”. Well, yes. I note that he mentioned one of my personal issues: the importance of data-centric applications architecture. I’ll be blogging more about this quite soon.
11:00am Ben Rushlo of Keynote on “ensuring high-performing, highly-available” cloud apps. Interesting choice of verb: “ensuring”. Developing? Operating? Hosting? Specifying? Hard to tell, because the audio has failed… The slides look as though they’re inspired by the various O’Reilly books on high performance and scalable web sites; stuff that I became very familiar with while at Amazon.
12:13pm Manfred Buchmann from SAP NetApp on storage infrastructure issues. Only just getting started… [...] OK, this is basically an account of how to put together Cisco, VMWare, and NetApp components to create a standardized, uniform, fully virtualized IaaS environment. Mostly a product pitch, albeit a very good one.
1:00pm Next up we have a panel on governance issues. Excellent set of speakers: Michelle Dennedy from Sun, Peggy Eisenhauer of PrivacyStudio.com, Malcolm Crompton of Information Integrity Solutions, and Nick Abrahams from Deacons (the last two from Australia). This panel nicely complemented the earlier session on legal issues: the earlier talk laid out the legal context, and these four panelists set forth the governance policies that companies need to adopt.
2:00pm Mark Masterson of CSC on Enterprise Cloud Computing. Starting off with a fairly idiosyncratic definition of clouds – specifically “Cloud” with a capital “C”, and “cloud” with a small “c”. Aargh!! After distracting us, he settles down to “cloud computing” as an architectural style for computing in elastic computing environments. Frightening the audience with over-elaborate architectural pictures, mind maps, and so forth which we are “not supposed to read”. Much of what he’s talking about has been addressed in Mark’s blog, which is worth subscribing to. Many of his slides include tons of footnote-style links to blog entries and other materials; in fact this is one of the most hypermedia-rich slide decks I’ve ever seen.
4:00pm Skipping one session (I’m not interested in learning “how to sell the cloud”), we pick up the conference with a talk on scalable storage by Boris Zuckermann, VP of Engineering at Ibrix. He’s going to be taking a file system centric approach, which is fine by me – I’m an old time NFS guy. He’s pitching the IBRIX “Segmented File System” as the right solution for the cloud. I asked the obvious question about the CAP theorem – consistency, availability, partition-tolerance, pick no more than two of three. In reply, Boris blustered that IBRIX “has an answer for all three”. I’m sure that Werner Vogels would be interested. [He's giving a keynote at this conference tomorrow at 7am Pacific.]
And I think that will do it for today. Still three days to go.
Comment from Mark Masterson
Time April 21, 2009 at 10:28 pm
Thanks for listening, and commenting, Geoff.
Glad you held on through the bits you didn’t like. FWIW, I get a lot of positive reactions to the trick (and it is certainly a trick) with the MindMap, so I’m of the impression that it’s a YMMV issue. And I did give you the URL to it on that slide as well, so you *can* go read it, if you want to.
If I succeeded in frightening people, however, then I’m happy with that.
Seriously, in an enterprise context (which was what I tried to frame my talk in), I think they should be at least worried. The argument for cloud computing is an economic one, but not strictly about cost reduction per se — a subtle distinction that escapes many people. And the truth is, current architectures will exploit elastic, parallel resources poorly — to the extent that this is true, that will mean inefficiency and wasted money. This is an elephant in the room (one of several) that needs to get called out, IMHO.
I’m really kind of passionate about the “big C vs. small c” distinction, so I’d be interested to hear more about why this got an “Aargh!” out of you? Although I acknowledge it’s a doomed proposition, I’d go so far as to say that, ideally, I wish we would all just talk (if at all) about the big C variant, and the small c variant would disappear forever. The common usage of things like “private cloud” (which most people really use to mean “internal cloud” — a system that is designed with remote resources attached to an enterprise’s network in a secure way sure seems to me to be a “private cloud”, despite the fact that a portion of it is external), for example, drive me batty.
But I’m happy to debate it with you, and I’m certainly open to being convinced to stop talking about it in this manner.
Cheers!