5.
Web Front Ends
All Application
Server Roles
SQL 2012
SQL Server installed and configured
to support SQL clustering, mirroring,
And AlwaysOn, AlwaysOn requires
SQL Server 2012.
Web Front Ends (2)
16 GB RAM
4 Cores CPU
100GB HDD (C:)
150GB HDD (E:)
All Application Server Roles (2)
16 GB RAM
4 Cores CPU
100GB HDD (C:)
150GB HDD (E:)
SQL Server 2012 (2)
32 GB RAM
8 Cores CPU
100GB HDD (C:)
500GB HDD (E:Data)
50GB HDD (F:Temp)
1.5TB HDD (G:Back)
SharePoint 2013 Production Environment
Web Front Ends
All Application
Server Roles
SQL 2012
SQL Server installed and configured
to support SQL clustering, mirroring,
And AlwaysOn, AlwaysOn requires
SQL Server 2012.
Web Front Ends (2)
4 Cores CPU
100GB HDD (C:)
150GB HDD (E:)
All Application Server Roles (2)
4 Cores CPU
100GB HDD (C:)
150GB HDD (E:)
SQL Server 2012 (2)
6 Cores CPU
100GB HDD (C:)
500GB HDD (E:Data)
50GB HDD (F:Temp)
1.5TB HDD (G:Back)
6.
Web Front Ends
All Application
Server Roles
SQL 2012
SQL Server installed and configured
to support SQL clustering, mirroring,
And AlwaysOn, AlwaysOn requires
SQL Server 2012.
Web Front Ends (2)
16 GB RAM
4 Cores CPU
100GB HDD (C:)
150GB HDD (E:)
All Application Server Roles (2)
16 GB RAM
4 Cores CPU
100GB HDD (C:)
150GB HDD (E:)
SQL Server 2012 (2)
32 GB RAM
6 Cores CPU
100GB HDD (C:)
500GB HDD (E:Data)
50GB HDD (F:Temp)
1.5TB HDD (G:Back)
Web Front End
All Application
Server Roles
Web Front End
8 GB RAM
4 Cores CPU
100GB HDD (C:)
150GB HDD (E:)
All Application Server Roles
12 GB RAM
4 Cores CPU
100GB HDD (C:)
150GB HDD (E:)
SQL Server 2012
24 GB RAM
6 Cores CPU
100GB HDD (C:)
500GB HDD (E:Data)
50GB HDD (F:Temp)
1.5TB HDD (G:Back)
SQL 2012
7. Operating System
• CPU – Over commit
• Memory – Dynamic?
• Network – IPV4
• Page Files – System Managed not
Auto
• Antivirus
• Disable Recycle Bin
7
8.
TCP 22233-22236 (Distributed
Cache)
TCP 808 (WCF)
TCP 5725 (FIM Sync)
TCP+UDP 389 (LDAP)
TCP 809 (Office Web Apps)
TCP+UDP 88 (Kerberos)
TCP+UDP 53 (DNS)
UDP 464 (Kerberos)
Make sure that the first server with
Distributed Cache has its firewall
rule set to allow for Inbound ICMP
(ICMPv4).
9.
10. SQL/SharePoint “Did you know’s”
SharePoint has its own Collation
Latin1_General_CI_AS_KS_WS!!
NTFS Allocation size (chkdsk) – SQL read and
writes 64k
How many SQL disks do you use?? Average?
Data/Log/Temp
Fix your RAM on your SQL boxes. Why?
11. Fill Factor. Indexing and data storage
performance – set to 80 to support growth
and minimize Fragmentation
All DB’s are created from MODEL – Set your
DB’s autogrow size correctly from the get go
Auto-create Statistics -- SharePoint does not
support it
SQL/SharePoint “Did you know’s”
12. SQL/SharePoint “Did you know’s”
Other options:
Parallelism – keep it at 1 for optimal query results
Instant File Initialization
tempdb size – Busiest DB so give it a decent initial size
(keep all autogrow the same across all settings)
Configure SharePoint to Autogrow after the DB’s have
been created
Shrinking Vs Backup Logs – Rather
backup and let it flush instead of shrink
13. SQL/SharePoint “Did you know’s”
Compress your backups
Recovery Mode – Simple
SharePoint logging grows exponentially
Create multiple Tempdb
files (.ndf) on multiple
drives or RAID 10 for drives
hosting Tempdb files
Defragment drives
containing content
database files
Don’t rebuild or manage
indexes of SharePoint DB’s
NTFS block size. Does
anyone care???
Hide SQL from other
computers
Disable Auto-Create
Statistics
Trace Flags
MODEL DB is set at 3MB. Always good to start with what you think you may hit in the first year. Log at 25%of the initial size
Parallelism – MAXDOP so that single SQL Server process serves each request
Instant File – Instant file initializes the first new space before it can be used, so it writes the DB with zero’s. But since 2005 it doesn’t so things happen faster.
Shrinking does not flush the DB and creates fragmentation
Parallelism – MAXDOP so that single SQL Server process serves each request
Instant File – Instant file initializes the first new space before it can be used, so it writes the DB with zero’s. But since 2005 it doesn’t so things happen faster.
Shrinking does not flush the DB and creates fragmentation