Why an OS
MY OS Notes
tuning & design
Support SAS Grid analytics
There are many suppliers. Also many suppliers connecting to SAS Grid can be found as technical papers or opinions.
Open source hadoop in-memory
, is in news as buzzing.
SAS support blog Troester
The GPU (instead CPU) could be used but not always. Some statements grid cuda Basel
As performance is very important to you, it should start with the hardware design in mind with the expected load. As we can´t foresee the future this will be continuously evolving process.
More a design question how to get the storage well behaved data sharing in grid
The balancing of the load is set up by a grid environment. This approach is completely different compared to a single machine or desktop usage thinking. See the paper:
391-2008 balancing load
Faq questions answers
The same picture is coming back in every grid analytics with SAS.
The following is the mindset to follow:
- More hardware is set up to get more performance. The Hardware components can be of a more common type.
Instead of more expensive hardware or dedicated machines to set up. This configuration is an dedicated cluster.
- The used hardware isn´t virtualized to save installed hardware. The virtualization is just an other managing layer
- The file systems with business data and code must be set up shared across all the related machines.
This mindset is complete different to thinking to save on hardware because of low impact loads to be shared on same machines.
The IO, the speed to the storage, is the most limiting resource in the building blocks as it is the slowest of all. So this has the most attention in speeding up.
Documents are found in the miner-7 system requirements.
- sgf2007 iosubsystem Best Practices for Configuring your IO Subsystem for SAS®9 Applications = (updated aug 2011, sastalks 2012)
- sgf09/270_2009 Maximizing the Performance of Your SAS® Solution: Case Studies in WebApplication Server Tuning for n-tier SAS Applications
Fail Save grid
Platform Ego is part of the Platform Suite for SAS. EGO (VMO) is a collection of cluster orchestration software components that, among other things, provide high availability to critical services.
The following figure shows the overall architecture and how these components fit within the SAS grid fail over context:
001-2009 High Availability
What is described is an advanced cluster of hardware setup with all options to deliver performance and availablity.
It is a dedicated setup, not one you would have imagined in this way.
It is the tooling of Platform that is used to solve all the requirements.
When checking this, it is IBM company (since october 2011).
is not implemented by SAS. They are focussing on big data.
it is possible ro use R by SAS/IML
In that way parallel processing for floating popint calculations can be optimized to higer limits.
Metadata server clustering
Enhance Your High Availability Story by Clustering Your SAS® Metadata Server in SAS® 9.4 .
Plan the hardware for expected usage
From SAS a view-point How to Maintain Happy SAS® Users
Virtualization in a UNIX Environment and SAS
is a paper describing the virtualization effects on hardware planning
is a paper with win2008 (win7) performance notes. The caching problem of large data is described
Left open the real question. Your datasize and work.
, are the key factors
on site = build your own saas
cloud or on-demand by saas delivered
Some Old links:
mp connect multi processing connect -----
desktop blog. VDI limiting in resources - performance
SAS Internal tuning
Reducing the CPU Time of Your SAS Jobs by More than 80%
When starting with analyses of SAS performance this is a good start practicalperf
employing host based tools
Options Tuning SAS
Transport - exchangeable
- ts140 the record layout of a data set in sas transport (xport) format
- 255-2012 The Use and Abuse of the Program Data Vector
the datasetstrcuture itself is propierty
SAS With OS tuning
Some SAS documents
Spread of SASwork was not easy within an Unix environment. With SAS 93 it can be configured to choose from multiple predefined SASwork definitions. Ask your sasdmin.
work round robin
Just in the process of sorting there is an third option aside the two of OS-level SAS performance, search for syncsort: 93 Config Unix
Some SAS documents
- 367-2012 A Hitchhiker´s guide for performance assessment & benchmarking SAS ® applications
- 72480_RHEL6_Tuning_Tips 2011 SAS® on Red Hat Enterprise Linux 6 (RHEL6) Tuning Guidelines
- 327-2009 Improving SAS I/O Throughput by Avoiding the Operating System File Cache
Some IBM documents
- WP101529 SAS AIX 5L, AIX 6 and AIX 7 Tuning Guide (Revised: 13-01-2013) SAS AIX .. Tuning Guide-2-pager.pdf
Abstract: These papers provide a starting point for performance optimization from a system-wide perspective to create an enhanced environment for SAS® 9 on IBM® POWER™ processor-based servers that run IBM AIX 6 and AIX 7.
- WP102277 SAS 9 on IBM Storwize family storage systems: Architecture, setup and tuning guidelines (Revised: 08-05-2013) SAS AIX .. Tuning Guide-2-pager.pdf
Abstract: This sheet provides a starting point for performance optimization from a storage architecture, setup, and tuning perspective to create an enhanced throughput environment for SAS® 9 on IBM® Storwize® family storage systems.
Mainframe SMF SDSF
whitepaper of SAS for getting Mainframe into the new approach. The TK service as scalability.
Mainframe SMF SDSF
The syncsort option mfx is an old option to boost sort performance mfx (syncsort mainframe)
SSIS SQL Server
A package is an organized collection of connections, control flow elements, data flow elements, event handlers, variables, parameters, and configurations, that you assemble using either the graphical design tools that SQL Server Integration Services provides, or build programmatically.
Integration Services (SSIS) Packages
Migrating DTS Packages to Integration Services
SQL Server Integration Services (SSIS) brings a revolutionary concept of enterprise-class ETL to the masses. The engine is robust enough to handle hundreds of millions of rows with ease, but is simple enough to let both developers and DBAs engineer an ETL process.
We Loaded 1TB in 30 Minutes with SSIS, and So Can You
This paper outlines what it took: the software, hardware, and configuration used. We will describe what we did to achieve that result, and offer suggestions for how to relate these techniques to typical scenarios.
Access 2010 specifications
This article has information about the limits of Microsoft Access database files and objects. In most cases when a database exceeds the following limits it might be an indication of a design issue.
Performance Tips To Speed Up Your Access 2007 Database
Some of the tips might contradict other tips, but they are offered because bottlenecks differ in each situation. Some tips might make things run faster on your system, while others degrade performance. You should evaluate each tip as it applies to your specific application running on your specific hardware.
Why an OS
MY OS Notes
© 2012 J.A.Karman (02 mar 2012)