Quantcast
Channel: SCN: Message List - SAP Adaptive Server Enterprise (SAP ASE) for Custom Applications
Viewing all articles
Browse latest Browse all 3587

Re: Process mode vs Threaded mode

$
0
0

Threaded mode is the default mode for 15.7 and newer for a reason, mostly to support today's modern multi-core, multi-threaded hardware.  So, one thing you need to consider is the hardware that you are running on and how many "engines" you currently run in process mode (hint: the fewer engines you run, the less "impact" i'll think you'll see from switching to threaded mode in general).   The other consideration is your workload.  Threaded mode provides a more efficient use of resources especially for workloads that require multiple active connections using disk, network, and cpu concurrently (which is a very general way to define a workload) on multiple engines.

 

There are great papers out there that compare and contrast both modes and discuss optimal configurations and workloads served by each.  If you're a DBA, it's your job to read everything available out there and make your own determinations based on your knowledge of your specific environment, from hardware, operating system, up to and including an understanding of how your business utilizes database resources (workload, transaction workflows), etc.

 

There is no generic "threaded/process mode is better" answer.

 

Filesystem based devices are fine as long as you understand how ASE interacts with them and optimize both the IO system, and ASE to use them most efficiently.  Also, it depends on what storage backing the filesystem are.  SAN?  SSD Flash drives?  SATA?   For Linux (depending on which version of REdHat you run), you need to pay attention to the configuration for "nr_requests" which controls an "io queue depth" of sorts for IO's to a filesystem.  Some versions of Redhat default to 128 which may be way too low for you.  If that queue depth is hit during a scheduled IO from ASE, then you effectively become blocked on that io and operate in a "syncronous" mode.  Obviously, you should disable journaling (never used a journaled filesystem for ASE storage), pay attention to block sizes, etc for all filesystems involved that host ASE  storage, or backup/recovery files.  There are also different IO schedulers in Linux as well.   To see which scheduler you are using:

 

cat /sys/block/sda/queue/scheduler

 

for example (to look at device "sda").   Read your OS docs about how to change schedulers and  which schedulers make most sense for your IO subsystem configurations .   This is all a large topic, and involves many layers (SAN, OS, Raid controllers, etc).  Again, you are the DBA and this is your job to acquire this information and apply it to your specific installation.

 

Good luck!


Viewing all articles
Browse latest Browse all 3587

Trending Articles



<script src="https://jsc.adskeeper.com/r/s/rssing.com.1596347.js" async> </script>