Comments on: Google’s Bigtable Distributed Storage System, Pt. I

By: jones

jones — Tue, 20 Feb 2007 22:28:34 +0000

gosh i wish this fiilesystem could keep up with novell nss

By: Robin Harris

Robin Harris — Thu, 08 Feb 2007 06:25:36 +0000

Warren Beatty not Steve Martin?
Halle Berry not Heather Graham?

Uh-huh.

But thanks for reading StorageMojo.

Robin

By: DAR

DAR — Thu, 08 Feb 2007 05:32:37 +0000

Sorry – that’s what I meant. Steve Martin, Eddie Murphy. I’ve seen it. Funny flick!

By: Robin Harris

Robin Harris — Thu, 08 Feb 2007 05:25:43 +0000

Bowfinger.

Robin

By: DAR

DAR — Thu, 08 Feb 2007 05:11:27 +0000

Great read. Thanks for the review!

Re: your trivia test: Bulworth?

By: Robin Harris

Robin Harris — Sat, 09 Sep 2006 21:38:03 +0000

Thanks for the links.

Sawzall – a great name – looks really interesting. Not sure I’ll dive into it in depth though. Here are a few thoughts while the topic is still fresh.

Sawzall works directly on the data in a Bigtable. Bigtable data is structured into tables, but is not typed. So it is up to Sawzall or any other app to know what it is looking at.

Ad hoc refers, I think, to a data set that is the product of a query or a Sawzall data reduction exercise. Once it’s created you can search it or perform more opearations on it.

I’m not quite sure what is meant by structuring search results on the input side. Page rank does just that, and there are allegations that Google plays with Page rank to provide results more to its liking, what ever that may mean.

A topic for a future post is “Finding meaning in massive data”. Search is the first step. Social networking, i.e. human-powered search is the second. Automating the discovery of meaning in massive data is, at least, the third.

By: e2eiod

e2eiod — Sat, 09 Sep 2006 17:36:02 +0000

Interesting read on “sawzall” at:
[Begin long URL]
http://www.networkworld.com/community/?q=node/8336&rlt=0904gibbs1&code=nlgibbs146124
[End long URL]

From the article:
Sawzall White Paper
http://labs.google.com/papers/sawzall-sciprog.pdf

And the movie mentioned in the article:
http://labs.google.com/papers/sawzall-20030814.gif

Structured, semi-structured and ad hoc Information spaces:

Sawzall on”ad hoc”
Traditional data processing is done by storing the information in a relational database and processing it with SQL queries. Our system has many differences. First, the data sets are usually too large to fit in a relational database; files are processed in situ rather than being imported into a database server. Also, there are no pre-computed tables or indices; instead the purpose of the system is to construct ad hoc tables and indices appropriate to the computation.

The words “ad hoc” do not appear in the Bigtable white paper.

Bigtable on structured, semi-structured Information:

“Bigtable is a distributed storage system for managing structured data that is designed to scale to a very large size:”

“Bigtable also treats data as uninterpreted strings, although clients often serialize various forms of structured and semi-structured data into these strings. Clients can control the locality of their data through careful
choices in their schemas.”

Does the Google search engine “structure” search results on the input side?
Is an effective search of “ad hoc” Information space not possible?
Is this search area size dependent?
Or is it resource limited?
Interesting…