SourceOne Journalling speed, throughput

(Lets try this reply again, I hit a shortcut key that submitted the response)

This response is not too scientific ,

I noticed that the journaling activities were not keeping up with the load of our journaling. 

I created multiple journaling mailboxes and spread journaling configuration of the mailbox databases between the multiple journaling mailboxes.  I then created multiple journaling activities so that it would process faster.

The back end needs to be able to support this increased amount of incoming data.

We went a step further and created separate mailbox databases for each country that we support, then journaling mailboxes for each one of those countries, and the journaling activities run separately for each country and place their archived data in separate native archives.  We can then implement separate retention policies for each native archive.

If you are new to SourceOne, they do have training courses that you can take remotely.  There is an administration one that is a week long.  It will show you how SourceOne functions, the terminologies, and how to plan a little (more planning is in another course i believe).  With that basic knowledge you can then apply it to your environment. 

That's not an unusual number as long it was taken into account for the overall design.

Can I assume the Journaling catches up off hours or does it continue to grow out of control?

How many S1 Servers are involved?

Physical or Virtual?

Any Idea who did the design and installation?

There are at a lot of factors that need to be looked at and incorporated into the design to properly support each environment.

After looking at CPU Utilization, Memory,Take a look at the Disk Performance on your S1 Storage Allocations.

Checking Avg Disk Queues, specifically the Message Center Drive but all the rest also, will give you an idea if your storage bound or not.

I would recommend having a S1 Professional take a look at and evaluate your environment along with bringing you up to speed on the basics of the product.

I would also highly recommend taking a SourceOne class available at education.emc.com if you are going to be maintaining this environment.

There are a couple online instructor lead training sessions coming up in December.

Start Date

End Date

Session

Location

Facility

12/10/2012

12/14/2012

M-F 9-5pm (CST)

Online - Americas

English

9 Seats Avail

12/03/2012

12/07/2012

M-F 9-5pm (Amsterdam, Berlin, Bern, Rome, Stockholm, Vienna)

Online - EMEA

English

4 Seats Avail

Gary Reardon

gCubed CTO

Information Governance and Archive Specialist

Consulting Infrastructure Solutions Architect

International Computerware, Inc. (ICI)

508.630.1604 413.967.4057

www.iciamerica.com

Gather, Guide, Govern

ICI Awarded:

EMC "Global Partner of the Year" for the 2nd Year in a Row"

EMC "Authorized Service Network (ASN) Quality" for the 5th Time"

VMware's "Business Continuity Partner of The Year 2012"

EMC IIG "Best Innovated Solution for Governance and Archive with gCubed Solution"

Thanks for the responses guys, its all useful info. I'm slowly reading through a Set of EMC Course notes and admin guide to see how it corresponds to our system. I may even get on a course sometime. I'm always grateful to use other peoples knowledge and experience but there's a lot to gained from breaking it and having to fix it yourself

The Journal mailbox does cleardown overnight.

We have a Physical SQL2008 server connected to a SAN that also holds the containers, The 'Master' server, and 2  worker servers are virtual servers. 1 worker server does most of the roles and the second one does Searches. Exchange2010 is running on a 2 server cluster with around 2500 mailboxes so not a huge system. In General the SourceOne servers seem very lightly loaded both in CPU and Disk activity. The Biggest CPU hog is an exasIdxobj.exe that can sometimes have 3 processes running taking up 80% CPU between them. Is that linked to SQL Indexing or is it Indexing the Containers or in fact doing something completely different ?

Greetings,

ExAsIdxObj.exe are the index threads responsible for creating full text indexes on file system and updating some information in SQL. Each server can run up to 4 such processes at a time as default. Out of the 4 processes you can allocate how you want processes to pick up data for indexing in the latest versions of SourceOne. See admin guide for more details around that part.

It is expected for ExAsIdxObj.exe threads to consume CPU. These threads come and go most of the time. You will be able to see the difference by turning on PID column in task manager. If a thread is found stuck for a very long time that is what you want to check if everyone is okay ? Majority of the times if these threads having any issue they will throw error messages in EMC event log on the server that are running.

Depending upon what is selected as options while running SourceOne web searches either SQL will be used or full text indexes from the file system will be used. There is separate set of binaries that will do the query work.

Thanks,

Rajan