Hardware for Scalix on ESXI + how fast should WAN link be?

[Uluquai]

Hi,

I'm considering moving a Scalix host to a different location.
2 questions:

Budget is not big, so I was thinking about putting it on a server running ESXI.

Would the following work out?

Serverhardware:
Bloomfield Xeon ~2,4GHz
8GB DDR3
4x 147GB 15K SAS Raid10 (exclusive to the Scalix quest)

..running 3 Quest OS's:
1. Scalix SBE on CentOS
2. Small Debian for some Network Monitoring
3. Windows Server running just a small mssql db + antivirus managment server, most services turned off.

maximum Scalix clients signed on concurrently: 15, ActiveSync: 5 clients

Would this setup work out well?

And about the Internet Connection:

4MBit/s up/down.. is this enough for a mailserver with max 300 mails/d (= mails in + mails out)?

Thank you very much!

[ls-al]

The hardware configuration sounds ok.
Re. WAN speed:
I have changed the Scalix installation for my company from a hosted box to an internally deployed system 2 months ago. We are on a 4 MBit SDSL too. Concurrent usage is a bit lower than in your case. But our traffic is in the area above one kilo per day.
We havent seen any problems so far. Sometimes SWA and even Outlook (non-cached) feels faster than with the hosted, dedicated server. And our hosting provider isnt too bad.

hope that helps

[Uluquai]

Thank you! That REALLY helps!

But what exactly do you mean by *sometimes* swa feels faster? Shouldn't it be smoking fast compared to an external host ? Or is your speed downstream so fast that you don't notice so much of a difference?At the moment (they haven't upgraded yet) their speed down is just 6mbit and that's really slow when opening other users mailboxes in Outlook...

Would you say that we could turn off smart cache when connected >100mbit to the server?

[ls-al]

Sorry. I wanted to say: "It feels faster when beeing accessed from _external_". For example when I read my mails from home.

The speed in the office is as expected. No need for smartcache with Outlook.

[Uluquai]

Ok.
But I think it's quite strange that when working externally you get faster access to the now "in-house" server with the rather slow uplink compared to when the server was hosted and probably was connected via 100 or even 1GBit/s.
Maybe latencies are more crucial?

[ls-al]

The latency is what we assume too.
But I did not have the time to check. Too busy with other things. Including posting here.