Google Gets Slack with Software Updates

An anonymous reader writes "While Google's open source project titled 'Slack' was released over a year ago, last week's Australian Unix Users Group Conference marks the first time that Google has ever discussed the system in public. Corporate systems administrator Michael Still helped to illuminate a little bit about how Google uses Slack and how their network of computers fits together. From the article: '"Slack is a source deployment system and it's the way we install applications on servers," Still said, adding Slack is based around a centralized configuration repository which is then deployed onto selected machines in a "pull" method. Each of the "worker" machines asks for its new configuration regularly or when a manual command is run.'"

Comments

Makes sense

[frosty_tsm]

Google has what, 700,000 servers? I would imagine that along the way they would have found existing solutions inadequate. Now they are making a version of their tool available other developers.

Re:Makes sense

[charlesnw]

1) The number of servers being used at any given time is pretty transparent. Do you think the average Google engineer has a way to do an actual count? Google's implementation of GFS (http://labs.google.com/papers/gfs.html) indicates that a user doesn't really have an idea of what server their using at what time... or if they're even using the same servers they were the day before. The actual filesystem is distributed and mirrored.

Well I imagine that accounting would have an accurate idea. And I was interviewing for a data center position. These people work with servers on a regular basis.

2) With the kind of growth that Google is known for (building of new datacenters, new offices, etc. all the time) the number is probably rapidly growing. From the time the engineer heard that unmber to the time when he leaked it to you to the day you're posting this, who knows where it has gone.

Very true. This was less then a year ago. But your right they do grow at a fair clip.

3) For what were you interviewing? Production? Corporate IT? Engineering? QA? What servers were you talking about? Do you think these are all sitting in the same place being used for the same things?

Corporate IT. I am not so stupid as to presume all the servers sit in place doing one thing. I work in a very large envrionment with multiple data centers and know that server counts are meaningless when you talk about things like virtualization etc.

Sounds similar to our setup

[nsanders]

We run Ubuntu in my department and ended up building an in-house Apt repository/svn/rsync system to maintain all our machines. We also use custom scripts that monitor NFS shares to emergency push operations. Obviously our down side is that an entire .deb package must be rebuilt for each change, but it's nice to see Google's method isn't out of this world after all.

Not new

[HavokDevNull]

Sounds very much like CFEngine http://www.cfengine.org/ with subversion?

radmind

[Anonymous Coward]

As soon as I read this I immediately though of radmind, which, by the vague descriptions seems to do exactly what is going on above. I encourage everyone to take a look!

Re:So its part of the Active Directory for Unix

[0racle]

No, its not possible. It's true (and I know) that OpenLDAP, Kerberos, Bind and a Network File system will give you a some of the functionality of the AD, if all you want is SSO and someway to centralize Automount settings and Printers. However doing DFS is not simple, pushing updates and revoking them is not as straight forward or robust, as the other poster pointed out and there is nothing like Group Policy.

This project seems to just provide another way to push applications to systems, presumably with the same limitation as any other, excepting maybe for Zen, where all your systems must be from the same vendor. It's from Google though so it must be gold.

how is that different/better than...

[oohshiny]

There are a bunch of widely used systems like this: rdist, cfengine, fai, ... In what way is "Slack" supposed to be better? Or is this simply a case of NIH?