Setting partition count globally

Hi,
I have the following questions about distributed services
1. I understand that the recommended value for partition count for distributed caches is the first prime number higher than the square of the number of JVMs. Is it possible to set this value globally for all the distributed services?
2. Are there any thumb rules for determining thread-count for my distributed services (for compute intensive and write-behind services)?
3. I have a large number of write-behind caches using using same distributed service. Is there any advantage of creating multiple distributed services for the same configuration and split up the caches between these services?
Thanks
Sairam
Edited by: SairamR on Aug 31, 2009 11:31 PM

Hi Magnus,
A few reasons for running multiple services are
- separation of code for testability (individual services can be more isolated from each other and could be developed by different teams, too)
- separation of services which would compete for resources, by having different layers of nodes, one layer storage-enabled for one service, another for another service
- separation of service threads due to reentrancy problems (cache store or other code running on the service thread of one service could this way use the other service)
- separating of data to a subset of nodes
You can't override partition count it on a global basis out-of-the-box, but there are several ways to do that
- either with appropriately changing the operational configuration file (tangosol-coherence.xml) adding a system-property attribute to the distributed cache service service-scheme part declaring the default value partition-count, or changing it to your liking
- or possibly (this may not work) also by defining an operational-config override file (overriding the respective element in the distributed cache service service-scheme part declaring the default value partition-count)
- or by having a common ancestor distributed-scheme for all the distributed cache services which defines a override with the system-property attribute
Best regards,
Robert

Similar Messages

Partition Count

Got couple of questions on the below thread!
Capacity 150 GB (actual Data) So ~3times 450GB total memory needed.
Out of this
150 GB = CacheName1 (application1 50GB) + CacheName2 for a specifc application (application2 100GB)
Question -
a) When I configure the partition count in the distributed scheme they should have separate partition count in their local-scheme for CacheName1 and CacheName2. (correct me incase)
PartitionCount Calulation:
50 GB actual Data (150GB x 1024) MB / 50 MB = 3072 ~= 3080 (3079 PrimeNumber + 1)
100 GB actual Data (300GB x 1024) MB / 50 MB = 6144 ~= 6152 (6151 PrimeNumber + 1)
b) Is there anyway to set operational parameter that reflects and restricts the CacheName1 to 150GB and CacheName2 to 300GB?
c) What is the role of High Unit(currently I have 2Gb for each of the CacheName1 and CacheName2 in the local scheme with all the JVMs running on 6GB)
My understanding is High Unit restricts the individual cache location like CacheName1 to not exceed 2GB in that node?
d) Does each cache name has some foot print in each JVM and is there any role by High Unit in the alloation of the cacheitems?

Hi
a) The partition count is preferably a prime number so do not add +1, the caches may have different partition counts but aren't required so
b) No not directly, since the capacity is dependent on the number of machines and their respective heap-size. If you had a stable environment you could use high-units to do something similar, by setting it to 300 (or 150) * 1073741824 / number of storage enabled jvms.
c) correct the high-units is what the local capacity of a certain scheme is.
d) if the hashCode() implementation have even enough distribution it should generate equal footprint for the machines in the cluster. However if you have data affinity or uneven distribution in the hashCode() implementation some partitions may become more heavy than others which if you are unlucky could end up on same machine.
Thanks
/Charlie Helin - Coherence Dev Team

Counting globals?

I have a question that is potentially difficult to put into
words.. but here goes!
I have a project, with 10 boxes for employee positions, and
15 boxes with employee names. What I want is a counter that shows a
number for how many positions are taken. At the moment, when an
employee name intersects a position, that name stays attached to
that position. So I could have a global saying that when an
employee intersects a position, set that global to true. But then
is there a piece of script that counts how many of these globals
are true, therefore giving me a number for how many positions are
taken? Or is there an easier way of doing this?
Thanks very much :)

This is a multi-part message in MIME format.
--------------040001080009090400040007
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
You could have another global which just counts the number of
positions
have been set. WHenever you add a persons name to a position,
increment
this new global by 1... if you remove a name, remove one.
--------------040001080009090400040007
Content-Type: message/rfc822;
name="Attached Message"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="Attached Message"
Path: forums.macromedia.com!not-for-mail
From: "Rich Bagnall" <[email protected]>
Newsgroups: macromedia.director.lingo
Subject: Counting globals?
Date: Tue, 3 Mar 2009 17:42:22 +0000 (UTC)
Distribution: macromedia
Message-ID: <gojq5u$c5f$[email protected]>
NNTP-Posting-Host: p01webapp11.macromedia.com
X-Trace: forums.macromedia.com 1236102142 12463 10.172.12.84
(3 Mar 2009 17:42:22 GMT)
X-Complaints-To: [email protected]
NNTP-Posting-Date: Tue, 3 Mar 2009 17:42:22 +0000 (UTC)
Xref: forums.macromedia.com macromedia.director.lingo:235977
I have a question that is potentially difficult to put into
words.. but here
goes!
I have a project, with 10 boxes for employee positions, and
15 boxes with
employee names. What I want is a counter that shows a number
for how many
positions are taken. At the moment, when an employee name
intersects a
position, that name stays attached to that position. So I
could have a global
saying that when an employee intersects a position, set that
global to true.
But then is there a piece of script that counts how many of
these globals are
true, therefore giving me a number for how many positions are
taken? Or is
there an easier way of doing this?
Thanks very much :)
--------------040001080009090400040007--

Setting Partition in U400

Hi everyone, I have just gotten U400.
Currently, first bootup: C: 470GB and D: 30GB, which is the Lenovo backup files.
Need help in setting partition in the C:. Is there any ways to add another partition?
I have used the Computer Management to set C: to 200GB. but it dont allow me to to set a "simple volume" on the unallocated space.
Is there any ways to setup the recovery to set the volume of the disk? I have done such for my pc using bootup.
but the U400 don't seem to allow me to do so
Will appreciated any advice

NitinJoshi wrote:
Thanks for the reply.
The create table statement above was just an example. Yes, i will go for YEAR interval.
But the thing is when i take Invoice_date in Prompt, I've to put Year(Invoice_date). Will that use partition or...?
Dear rp0428,
my queries would be normal select from Invoice Sales table but where clause would have year,organization_descr and 1 more column. year & organization_descr filter would remain for most of the analysis.
Regards!
We aren't asking for some general statement about what your query 'might be'; Oracle works with actual queries. If you want an 'actual' answer provide an 'actual' query.
Oracle can't use 'year' to query a column of type DATE unless you turn it into a date using TO_DATE.
And the Oracle server knows NOTHING about 'Prompt' or 'Year(Invoice_date)'.
Partition pruning can only occur if the partition columns are included in the WHERE clause.

How to set Tile count column or count row?

How to set Tile count column or count row? If not can do, How
to adjust count row or count column?
Thx for all idea.

The Tile container's number of columns and rows is calculated
based on each child's width and height (or based on tileWidth and
tileHeight which you can set). If you need more control over the
layout, use Grid.

Table partitioning and global indexing

hello all,
the requirement is to do table partitioning with global indexing! & also thr r almost 60 crores of rows are there! so can you tell me the way which can be done better and also please provide me the procedure with documents?
thanks in advance!

Hi,
have a look at online redefinition of a table, see Oracle Manual (10.2): http://download.oracle.com/docs/cd/B19306_01/server.102/b14231/tables.htm#ADMIN01514
The steps are described in the manual.
Herald ten Dam
Superconsult.nl

Set image counter to 0

hi,
I have Nokia 3230 mobile phone, i'm going to remove all image from memory card and set image counter to 0, do you know how can i it do?

Hi aghaghi
You could try reset *#7370# after backing up essential data.
Also you may need to reformat your memory card.
Happy to have helped forum in a small way with a Support Ratio = 37.0

Create all jobs in own partition or global

Dear Friends,
I have created own partition in production to schedule all jobs and using GLOBAL partition in QA system.
I can not do Import in own partition from GLOBAL(QA system) partition.
My concern is whether we can perform all activities in own partition same as GLOBAL partition.
OR we need to use GLOBAL partition in production instead of own partition.
Any help is highly appreciated.
Thanks in advance.
Regards
Jiggi

Dear Blom,
Thanks a lot for your helpful answer.
In that case, will change QA partition from Global to Own same as Production.
Regards,
Jiggi

What is the difference between partition-count and the number of caches?

What is the difference between partition-count and the number of caches in Coherence? Are they same?

Those are totally orthogonal concepts.
For more, look at this thread where I answered your other related questions and explain this, too:
Where can I find the accurate definitions of these terms?
Best regards,
Robert

Setting up a global forward

Please let me know whether it is possible to set up a global application such that any request coming into the server will hit the global war before it hits any other deployed application. The deployed applications need not be aware of the global application.

I've thought about quicksilver, but i understand that's it's used to do much more than just that. And, I don't really want to deal with all of quicksilver. Now that might change, but I think I want to stay away from it for now.
However, butler looks good. I'll play around with it.
Also, the other application "Fast Scripts" looks decent, but it's shareware and I personally would like to get an entire program free and then donate when I feel props are due. (Just my personal thing.)

Good partition-count for 360GB cache

Coherence Version: 3.6.1.3
JRE Version: 6u22
Platform: Solaris 10
We have a application which uses a 360GB coherence cache with 60 cache nodes. What is good partition-count configuration. The current coherence count we are using is 3607, but we are seeing some issues with query functionality. So I check the Oracle docs to see if this partition-count is a problem but its not helping. They have 2 points:
1. The number of partitions should be a prime number and sufficiently large such that a given partition is expected to be no larger than 50MB in size
2. They have given table and say that for 100GB the partition count should 8191
If point #1 is considered the partition-count of 8191 is too high for 100G node.
Edited by: coh on Apr 16, 2012 11:54 AM

Hi
Whilst NJ is right and GC should be tuned you should also look at why you might have so much GC - the chances are that the queries are producing the high GC rather than the GC affecting your query. We had a situation recently where the general performance of our system had halved and this turned out to be a simple POF extractor Filter query that was running quite frequently and was not using indexes. As this was in effect a full scan of one of our biggest caches then a lot of garbage was generated. In our case the GCs were small as the garbage was in the new generation, but even so this had quite an impact on the system.
What sort of queries are you doing? Are they using reflection extractors or POF extractors? Do you have indexes?
user738616 wrote:
With your configuration, you are looking close to 50MB/paritition and a total of 3-4 GB data transfer assuming only one of the nodes was running stop-the-world GC.NJ: I don't quite understand what you are on about here - a stop the world GC is not a cause of partition transfer. Partition distribution will only occur when nodes join or leave the cluster. If you are seeing partition transfer during queries then you may have killed one of the nodes or you caused a GC that was so long that the node left the cluster. Long GC pauses will cause partition transfer to take longer as the nodes that need to participate in the distribution are paused.
JK

Setting Partition Arguments (log flags) using FSCRIPT??????

Hi,
I can't succeed in setting parameters in FSCRIPT that require quotes (as
__-fl "(err:sh:*)"__:
1)
SetPartArgs VisionFrontEnd "-fl aFile(err:sh:*) +(trc:user9:*:*) -fm
(n:1024,x:800000,i:128)"
This sets the line, but the partition can't start.
2)
SetPartArgs VisionFrontEnd "-fl "aFile(err:sh:*)"
"+(trc:user9:*:*)" -fm "(n:1024,x:800000,i:128)"
(with or without '\' protecting)
This sets the line to 'empty', with an error saying that aFile(err:sh:*)
can't be found
How to set something on the parameter arguments list that contains log flags
?

While the user long and short name is a matter of convenience when logging in (the computer will "prefill" the user name at the server login window based on what account was used to log on to the local machine), what counts is the UID number of the account.
Most "home" user accounts end up with a "501" UID meaning that it is the first account created and it is an administrator account on the home machine.
What we have found is users are quite creative as to what they use for account names on their home machines and most times they are not compatible with the company standard of first initial last name we use for accounts.
So we let them do what they want but let them know the login and password of the network account so they receive access to network resources based on that account.

A few questions on setting partitioning the hardrive.

Hello,
I am reinstalling Arch because I want to make the jump from i686 to Arch64. I really want this to be the last time I reinstall an OS on this laptop, so I have a few questions on the most efficient way to set up the partitions.
1) Is it safer to have a seperate boot partition? I really don't mind setting aside 100MB or so for boot, but I keep hearing you really don't have to do that with Arch anymore.
2) How does optimization work? Are partitions that are created higher on the list faster to access? What I'm looking for is a really fast computer while in use. I don't care about boot time very much, and my swap drive is barely used. Would this be a good set up for what I'm looking for?
SDA1 Root
SDA2 VAR
SDA3 Home
SDA4 Swap
SDA5 Boot
3) Is ext4 a good idea? I really like the speed and all that, but it's still in development, right? When it's finished, would I be able to update the filesystems or once I make them are they always stuck in their current state? I was thinking ext2 for boot then ext4 for everything else.

I'm not too sure on that one as i have never tried it but it might be possible to boot from your live cd, backup your /var partition reformat it then copy the data back on to it but don't quote me on that.
I don't think you will notice any difference regarding which order you mount your partition's, I have been through a bunch of different partition setup's and have not noticed any speed difference between them only between different filesystem's
I have the same partition setup as your first post minus the boot partition
/dev/sda1 / ext4 defaults,noatime 0 1
/dev/sda2 /var reiserfs defaults,noatime 0 1
/dev/sda3 /home ext4 defaults,noatime 0 1
/dev/sda4 swap swap defaults 0 0

How to group result set to count values in several (2) attributes?

Hi there,
the (simplified) query at the bottom returns the following result set:
OBJ_ID Attr 1 Attr2
22674886 HAK44221 GEB132542
22674886 HAK44221 GEB92751
22674886     HAK44222 GEB92744
22674886     HAK17113 null
However, we need the number of different Attr 1 and Attr2 counted, which should result in the this:
OBJ_ID Attr 1 Attr2
22674886 3 3
We trried using "group by grouping sets", "group by" and also took a glance at "cube" but didn't manage to get the desired result, mainly due to the fact, that Attr2 may contain null values.
We managed to find a solution using a GTT or a view. Can anybody give a hint how to solve that without using temporary tables?
Best regards
Pat
SELECT
o.obj_id,
o2.obj_name as Attr1,
g.ckw_nis_nummer as Attr2
FROM objekte o
JOIN eigenschaftsdaten d0
ON (o.obj_id = d0.obj_id AND d0.eig_id = 2525755)
LEFT JOIN eigenschaftsdaten d8
ON (d0.egd_wert=substr(d8.egd_wert,1,instr(d8.egd_wert, ' ')-1) and d8.eig_id=2525976)
LEFT JOIN objekte o2
ON o2.obj_id=d8.obj_id
LEFT join gebaeude g
ON o2.obj_id=g.parent_house_service

Hi,
That looks like COUNT (DISTINCT ...):
SELECT       o.obj_id
,       COUNT (DISTINCT o.obj_name)          AS attr1
,       COUNT (DISTINCT g.ckw_nis_nummer)     AS attr2
FROM        objekte o
JOIN        eigenschaftsdaten     d0     ON   o.obj_id      = d0.obj_id
                           AND d0.eig_id       = 2525755
LEFT JOIN eigenschaftsdaten      d8     ON   d0.egd_wert = SUBSTR ( d8.egd_wert
                                                         , 1
                                          , INSTR (d8.egd_wert, ' ') - 1
                         and d8.eig_id=2525976)
LEFT JOIN objekte          o2     ON o2.obj_id      = d8.obj_id
LEFT JOIN gebaeude           g     ON o2.obj_id      = g.parent_house_service
Depending on your data and your requirements, you may need to join some of the tables separatemly, do a GROUP BY to get attr2, and then join those results to the other tables.
I hope this answers your question.
If not, post a little sample data (CREATE TABLE and INSERT statements, relevant columns only) for all the tables involved, and the results you want from that data. WSimplify the problem as much as possible. For exam[ple, if your real problem involves 5 tables, but the part you don't understand can be shown with only 1 or 2 tables, then post sample data and results for those 1 or 2 tables only.
Point out where the query above is getting the wrong results, and explain, using specific examples, how you get those results from the sample data in those places.
Always say what version of Oracle you're using.
See the forum FAQ {message:id=9360002}

About: making RAID sets - partition independant?

(the querry made simple: look last 2 lines)
When making a RAID set, whatever the type it is, will it rely on the partitions or on the disk itself?
In the Disk Utility help it keeps mentioning "disk", and when I was about to create a set, it also asked me about "disk".
I have set up 2 partitions for the purpose of testing RAID for me, being about the same size on a big backup drive.
The drive itself also has a bit of free space filled with original files.
So I wanted to mirror, say, RAID test 1 and RAID test 2, both of them being on the same disk.
Will it work?
Or will it wipe my disk?

A RAID may be constructed using individual partitions on separate drives although using the whole drive would be preferred for speed. You could actually RAID multiple partitions on the same drive, but of course what would be the point.
I've actually used two drives each with three separate partitions and created three separate RAIDs of two partitions each. Not particularly efficient but it does work.
There is really no point to creating a RAID of two partitions on the same drive. And, certainly no point creating a mirrored RAID this way.
Any RAID you create with Disk Utility will erase the drives used (or partitions.)
RAID Basics
For basic definitions and discussion of what a RAID is and the different types of RAIDs see RAIDs. Additional discussions plus advantages and disadvantages of RAIDs and different RAID arrays see:
RAID Tutorial
RAID Array and Server: Hardware and Service Comparison>.
Hardware or Software RAID?
RAID Hardware Vs RAID Software - What is your best option?
RAID is a method of combining multiple disk drives into a single entity in order to improve the overall performance and reliability of your system. The different options for combining the disks are referred to as RAID levels. There are several different levels of RAID available depending on the needs of your system. One of the options available to you is whether you should use a Hardware RAID solution or a Software RAID solution.
RAID Hardware is always a disk controller to which you can cable up the disk drives. RAID Software is a set of kernel modules coupled together with management utilities that implement RAID in Software and require no additional hardware.
Pros and cons
Software RAID is more flexible than Hardware RAID. Software RAID is also considerably less expensive. On the other hand, a Software RAID system requires more CPU cycles and power to run well than a comparable Hardware RAID System. Also, because Software RAID operates on a partition by partition basis where a number of individual disk partitions are grouped together as opposed to Hardware RAID systems which generally group together entire disk drives, Software RAID tends be slightly more complicated to run. This is because it has more available configurations and options. An added benefit to the slightly more expensive Hardware RAID solution is that many Hardware RAID systems incorporate features that are specialized for optimizing the performance of your system.
For more detailed information on the differences between Software RAID and Hardware RAID you may want to read: Hardware RAID vs. Software RAID: Which Implementation is Best for my Application?

Setting partition count globally

Similar Messages

Maybe you are looking for