Testing tools and AI - ideas what to try with some tool examples
Avoiding 10 common SharePoint Administration mistakes
1. Dodge the
bullet
10 ways to avoid common
mistakes in SharePoint
Administration
#SPSUK08
Benjamin Athawes
www.benjaminathawes.com
@benjaminathawes
2. Who am I?
− SharePoint server junkie for
just over 3 years
− Head of IT Infrastructure at
eShare Limited in West
Berkshire (near Reading)
− SharePoint 2010 admin/dev
MCTS, MCITP & MOS
− SharePoint User Group
speaker (#SUGUK)
2
3. The slide my boss
made me add…
• The only specialist provider of “governance” software
• 220 + trusts
• Pensions / Health / Charities / Corporates
• UK / Holland / Australia / South Africa
• Hosted solution based on SharePoint Server and K2
• Data centre located in Newbury, Berkshire
5. Who are you?
• IT PRO?
• Dev?
• Business?
• A bit of everything?
• Wrong session?
6. Why this session?
Similar administration questions keep recurring with each
new version of SharePoint...
“I’ve made changes
in <SQL/IIS/the file “I have 100 Web apps
system>, why have I and my farm seems
got “an unexpected slow. How do I speed
error”? things up?”
“I have a 500GB
content database and “Why am I getting a
I’m having trouble warning in Central
upgrading to Administration saying
SharePoint <version>, that the Farm Account
what can I do?” shouldn’t be a local
admin?”
6
7. Why the same old questions?
• Sophisticated product
• Poor documentation
• Success involves
collaboration
• Growing complexity!
9. 1. Put SharePoint in your job title
− BIG product - not just
SharePoint
− „One man army‟ probably
not realistic
− Training
10. 2. Start with an appropriate topology
Yup,
− Beware of SharePoint‟s another
default settings: standalone
install
SQL Server Express
(R2 =10GB/DB limit)
No User Profile Service!
− Plan the install
11. The evil button
SQL Server Express
Full fat SQL – use this
12. 3. Understand capacity planning
recommendations
− Easy to push under the rug
− Software boundaries normally
involve a trade-off
− No magic numbers or a silver
bullet (e.g. RBS)
− Not always documented e.g.
Web application scalability
13. Capacity Planning Example: too many
Web apps
Web Apps
Department Department Public Site A Project A Project B
A B
Collections
Department Department Public Site A Project A Project B
A B
Site
Databases
Content
Department Department Public Site A Project A Project B
A B
…n Web apps; up to a recommended maximum of 20/farm
14. Capacity Planning Example: scaling
with site collections
Web Apps
Secure Web Public Web SPCA MySites CTHub
App (HTTPS) Sites (HTTP)
Collections
A B C D E A B C D E A B C D E
F G H I J F G H I J SPCA F G H I J CTHub
K L M N O K L M N O K L M N O
Site
P Q . . . P Q . . . P Q . . .
Databases
Content
Secure Web Public Web SPCA MySites CTHub
App (HTTPS) Sites (HTTP)
…n site collections; recommended max 2000/content DB (!)
15. Common reasons for Web apps…
Reason Potential Solution
Vanity URLs Host named site collections
Extend Web app (2007); multiple providers in 2010
Authentication provider
in single Web app
SSL Extend Web app
Partitioning Multi-tenancy (not trivial to implement)
Web.config customisations Try to avoid them!
16. 4. Utilise SharePoint's provisioning
capability
− If it‟s not packaged in a
solution, don‟t deploy it!
− Solutions are:
− Scalable
− Backed up
− Quicker (e.g. branding
artefacts on file system)
18. 5. Avoid the default SQL DB growth
settings − Default SQL autogrowth settings
mean fragmentation
Pre size ldf/mdf files
Autogrow in MB not %
− Index maintenance still
relevant to SSDs
− Mechanical disks – don‟t forget
OS fragmentation!
19. 6. Remember that virtualisation isn’t
magic sauce
− You can‟t get away from
normal physical limitations:
− SQL IOPS
− Hardware failure
− NIC throughput
− Ensure Non Uniform Memory
Access (NUMA) has a direct
impact on:
− RAM installed in box
− Allocation of RAM to VMs
21. The NUMA misnomer?
− NUMA = a scalable CPU design
for multi core processors
(a good thing).
− The “rule of thumb” guidance for
determining NUMA boundaries:
− RAM / logical cores
− E.g. 64 GB RAM / [2 * quad
core CPU] = 8GB boundary
− My guidance: “It depends” on
your hardware:
check your specific configuration
using CoreInfo & PerfMon
− Hypervisor agnostic
22. NUMA example
A HP DL380 G7 with 64GB RAM and 2 hex core processors
Hyper threading enabled
In this particular case, there is a 1:1 ratio between CPU sockets and NUMA nodes,
meaning that there are 2 NUMA nodes of 32 GB each.
23. 7. Design and test a back up strategy
− Align strategy with SLAs
− Remember SQL backups
don‟t cover everything (farm
config)
− Document your farm
configuration using
PowerShell or 3rd party tools
e.g. SPDocKit
24. What’s backed up?
Configuration Setting 2007 SP2010 farm-only config SP2010 with SPDocKit
A/V settings. N Y Y
Diagnostic logging settings. N Y Y
IRM N Y Y
Farm solution store N Y Y
Application pool settings, including service accounts (all
accounts that run as Web applications, including the crawler N N
account and the search account). Y
Alternate access mapping settings. N N Y
Farm-level search settings. N N Y
External service connection settings. N N Y
Workflow management settings. N N Y
E-mail settings. N Outbound only Y
Usage analysis processing settings. N N Y
Content deployment settings. N N Y
Timer job settings. N N Y
Recycle Bin settings and other Web application general
N N
settings. Y
Administrator-deployed form templates. N N Y
Default quota templates. N N Y
Database names and locations. N N Y
Web application names and databases. Be sure to document
the content database names associated with each Web N N
application. Y
Crawler impact rules. N N Y
Activated features. N N Y
Blocked file types. N N Y
Check out SPDocKit at http://tinyurl.com/c6p6lus
25. 8. Test, test and test again
“No test farm equals no
production farm!”
Andrew Woodward
− Includes:
Cumulative updates
Service Packs
Topology changes
IIS changes
− Every environment is
different
26. 9. Utilise PowerShell
− STSADM is deprecated
− Not just for developers
− MS product agnostic
− Dozens of uses: removing
GUIDs, automating installs,
log checking, etc
…but don’t forget to test your
scripts!
27. 10. Use crowd wisdom
− “Many Are Smarter Than the Few”
− Use social networking tools to keep up to date, e.g.
“RBS is not „the solution to world hunger‟”
Spencer Harbar
“Cumulative Updates are not Service packs”
Todd Klindt
“SharePoint‟s farm-only config leaves a lot to be desired”
Sean McDonough
28. Summary
− It‟s not just SharePoint
− Understand the capabilities and supported scenarios
− “Default settings” aren‟t always sufficient
− Social tools can aid decision making
− Plan early to “dodge the bullet”
28
29. Thanks…
…to you for listening
…to Sean McDonough
for reviewing most of the
slides
Notas do Editor
Google+ also offers Hangouts - a free video conferencing tool for users. We included Hangouts as we see video being a powerful way to connect users.You choose who to hang out with, up to a maximum of ten people. If you want more exposure you can broadcast your hangout to the web; you can even watch a YouTube video together and share documents.Michael Dell is already using it to talk to his customers, TV presentersin the US are using it to talk more intimately with their audiences, and Will.i.am is using as part of his engagement during concerts. The Dalai Lama is also using Hangouts.You could imagine a time when Landrover runs a competition for 10 enthusiasts to hangout with the engineers behind the car; and the rest can just watch the broadcast
Good afternoon, it’s great to be here as part of Mindshare’s Social Immersion Day. My name is Paul Coffey and I am one of the leads for Google’s social advertising products in the UK.
Default settings:Data: 1MBLogs: 10%“While SSD’s might reduce the IO latency for random read operations, they don’t negate the need to continue to adhere to proper design principals for primary keys, fill factor on indexes, and index maintenance. ”“Paul’s answer to the question was very practical and dealt with the wasted space utilization that excessive index fragmentation can cause in a database that uses a uniqueidentifier column with newid() as the primary key, and given the cost per gigabyte for SSD storage this could be quite significant.”“One thing to consider is that an SSD while faster still has an upper limit to the number of I/O operations per second (IOPS) that it can perform. ”http://www.sqlskills.com/blogs/jonathan/post/Does-Index-Fragmentation-Matter-with-SSDe28099s.aspx
(KB 841057) Content DBs fully normalised for scale / efficiency
(KB 841057) Content DBs fully normalised for scale / efficiency
(KB 841057) Content DBs fully normalised for scale / efficiency
(KB 841057) Content DBs fully normalised for scale / efficiency
RPE = “Brian Alderman” from Best Practices 2011 in London