Transcript Top 10 non-SharePoint Technical Issues that can doom your

What You’ll See
• Slow performance across the
board
• Long search crawl times
What’s Wrong
• SQL Server is the bottleneck in
SharePoint implementations
What to Check
• SQL Server Buffer Manager: Page
Life Expectancy > 300 (seconds)
What to do
• Add memory
• Add memory
• Improve disk performance
• Add another SQL instance
Platte
r
Arm
Head
RAID 1+0
RAID 5
A1a
B1a
C1a
D1a
A1b
B1b
C1b
D1b
A2a
B2a
C2a
D2a
A2b
B2b
C2b
D2b
Disk 0
Disk 1
Disk 2
Disk 3
A1
B1
C1
Dp
A2
B2
Cp
D1
A3
Bp
C2
D2
Ap
B3
C3
D3
Disk 0
Disk 1
Disk 2
Disk 3
What You’ll See
• Slow performance, especially
initial loading of a site
• Less effectiveness in adding RAM
to the SQL server
What’s Wrong
• Disks aren’t producing enough
IOPS
What to Check
• Average time per read < 20ms
• Average time per write < 20ms
• RAID configuration is RAID 10 not
5/6 (DP)
What to do
• Reconfigure from RAID 5/6(DP) to
RAID 10
• Buy more, faster, smaller disks
What You’ll See
• First hit response times, even
after the first hit
What’s Wrong
• The load balancer isn’t configured
for session affinity (sticky
sessions)
What to Check
• Session affinity
• Session identification
What to do
• Configure the load balancer for
session affinity (sticky sessions)
What You’ll See
• SharePoint Server disks get filled
up
What’s Wrong
• The full text (non-SQL) indexes
get copied to every query server
(and they’re big)
What to Check
• Every query server has 30% of the
total size of everything you want
to search
What to do
• Add disks until you have enough
What You’ll See
• Weird SharePoint errors
• Erratic and unpredictable
performance
What’s Wrong
• Insufficient bandwidth between
servers
• Packet loss
What to Check
• No Firewall between servers
• Network adapter transmit buffer
<1
What to do
• Remove firewall and proxy
servers
• Ensure all network ports plugged
in and bonded
• Plug all SharePoint servers into a
single switch
What You’ll See
• Unknown problems in your
production environment
What’s Wrong
• There’s no way to validate your
code and configuration changes
What to Check
• The value of downtime in
production to the organization
What to do
• Get at least one controlled
environment before production
What You’ll See
• Random outages in production
• Random performance issues in
production
What’s Wrong
• Production is referencing
development resources (which
are being shutdown)
What to Check
• Production can not communicate
with development
What to do
• Put production (and QA) on a
separate subnet that can’t talk to
development resources
What You’ll See
• All clients look like they’re
coming from one IP address
What’s Wrong
• The proxy server or load balancer
is NATing (abstracting) the IP
address
• Difficult to diagnose specific
client problems when you can’t
find them
What to Check
• Configuration of the proxy server
or load balancer
• IIS Logs to verify client IP
addresses are being recorded
What to do
• Reconfigure the network to
transmit client IP addresses to the
server
What You’ll See
• Nothing
What’s Wrong
• You can’t fix what you can’t find
What to Check
• Monitoring options available to
you
What to do
• Install monitoring
• Configure monitoring
What You’ll See
• Prompts for login (like every 30
seconds)
What’s Wrong
• IE with your web site not in the
Intranet zone
What to Check
• IE settings
• Time synchronization
What to do
• Set a group policy
• Make workstations a member of
the domain
http://channel9.msdn.com/Events/TechEd
www.microsoft.com/learning
http://microsoft.com/technet
http://microsoft.com/msdn