[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Poll: versions and setups in use



Hi

We use the version (Release 7.2.6_GA_2926.UBUNTU8_64 UBUNTU8_64 FOSS edition) running on a XenServer machine Virtual 1 processor com 8 cores e  16 GB of memory
We have accounts in 1400.
MX and smtp run on separate virtual machines with 2 GB of memory.


Greco
SP-Brazil


De: "Steve Hillman" <hillman@sfu.ca>
Para: "Enrico Weigelt, metux IT consult" <enrico.weigelt@gr13.net>
Cc: zimbra-hied-admins@sfu.ca
Enviadas: Terça-feira, 23 de setembro de 2014 13:17:14
Assunto: Re: Poll: versions and setups in use

At SFU, we're on version 8.0.7, but we only just upgraded a few weeks ago from 7.2.6. The upgrade process started in March. I'd be surprised if anyone on this list has upgraded to 8.5. Upgrading to a .0 release is always risky, and for a site of any significant size, the upgrade project typically takes weeks, or even months, to work through.

We run nearly 70,000 users and another 40,000+ distribution lists, though we only use the DLs for ACLs within Zimbra - actual delivery happens on external MTAs. Our MTA/Proxy/LDAP servers are virtualized (3 of them), while the mailbox servers are still physical. We run 4 mb servers, with about 15-18000 users on each. Each server has 24 cores and 192gb ram, so it's a bit big to be virtualized. I doubt we'll look at virtualizing our mailbox servers until we make them smaller, which we may start doing prior to an 8.5 rollout. Total storage is in the ballpark now of around 30TB. All users get an initial 5GB quota, but faculty/staff get theirs automatically bumped when they get to 90% in-use.

We've been on Zimbra now since version 5.0.11 - early 2008. We run the Network Edition (again, you'll find almost everyone else on this list does too). The price we pay for professional support and extra features is tiny compared to many of our other enterprise licenses. The main NE feature we use is ActiveSync, which we make extensive use of. We don't use Zimbra backups -- they don't scale well to 10's of thousands of users. Instead we use NetApp back-end storage and take regular snapshots of LUNs. On a weekly basis, we auto-create a staging environment based on those snapshots. That serves two purposes -- it continually verifies that our snapshot process is sound as a means of creating a working backup, and it provides us with a backup environment should we need to do a restore for a particular account. Since the introduction of the Dumpster though, requests for restores have dropped to near zero. I don't actually remember the last time we did one.

When we first were thinking about deploying Zimbra to replace our open-source/in-house mail system, we really wanted to run it on Solaris. At the time, Solaris was still fairly dominant in enterprise data centres and had lots of superior features. Zimbra wouldn't support Solaris, so we played around with getting it to work on Solaris ourselves and considered running the open-source version. In the end though, we decided it would be too difficult to support long-term, and went with Linux (it was our first enterprise application to be hosted on Linux). Since then of course, Oracle bought Sun, Solaris has withered, Linux has taken over, and virtualization has rendered many of the Solaris-specific features moot

----- Original Message -----
> On 10.09.2014 00:30, Dusko Pijetlovic wrote:
>
> Hi,
>
> > Our Zimbra version is Zimbra 7.1.4_GA_2555 (build 20120104215751).
> >
> > Number of accounts: 455
> > Number of distribution lists: 70
> > Number of resources: 42
>
> thanks.
>
> Is there any special reason for still running that old version
> (not even upgrading to recent HELIX release) ?
>
> > It runs on Red Hat Enterprise Linux Server release 5.10 VM with 4 CPUs,
> > 24 GB RAM, 1.9 TB storage space -- It just works (tm) for us.
>
> Physical or virtualized ?
>
> In the last decade, I really became a big fan of virtualizing
> everything. Container solutions like OVZ or LXC make the extra
> cost practically zero, but it really makes dailly operating easier
> (eg. when the system has to be moved to another hardware node)
>
> > I'm interested in the results you collect and am wondering if you can
> > send them to me once you compile them?
>
> Yes, I'll do so. For now, you're the first and only one :)
>
>
> By the way: anyone here using the NE ?
>
>
> mit freundlichen Grüßen
> --
> Enrico Weigelt,
> metux IT consulting
> +49-151-27565287
>

--
Steve Hillman        IT Architect
hillman@sfu.ca       IT Services
778-782-3960         Simon Fraser University