Max number of instances per server

Hey all, I am looking into the feasibility of using HDR from 2000 sites to a central server. Is it possible for a single server to hold 2000 instances, and has anyone else out there done that? Any details you have on how you run queries against that as well would be great info. -- Will chat again soon, Kate katedarts@gmail.com --001a1144491cc9cb96053f3ab9e0

To query all 2,000 instances you would have to connect them in a GRID and perform a gridded query across the entire grid. My record is 10 server instances on a single system. Art Art S. Kagel, President and Principal Consultant ASK Database Management www.askdbmgt.com Blog: http://informix-myview.blogspot.com/ Disclaimer: Please keep in mind that my own opinions are my own opinions and do not reflect on the IIUG, nor any other organization with which I am associated either explicitly, implicitly, or by inference. Neither do those opinions reflect those of other individuals affiliated with any entity with which I am affiliated nor those of the entities themselves. On Wed, Oct 19, 2016 at 12:59 PM, Kate Tomchik <katedarts@gmail.com> wrote: > Hey all, > I am looking into the feasibility of using HDR from 2000 sites to a central > server. Is it possible for a single server to hold 2000 instances, and has > anyone else out there done that? Any details you have on how you run > queries against that as well would be great info. > > -- > Will chat again soon, > Kate > katedarts@gmail.com > > --001a1144491cc9cb96053f3ab9e0 > > > ************************************************************ > ******************* > Forum Note: Use "Reply" to post a response in the discussion forum. > > --001a114a108040117e053f3acf9f

255: http://www.ibm.com/support/knowledgecenter/SSGU8G_12.1.0/com.ibm.adref.doc/ids_a dr_0718.htm If you have a server big enough you can create VMs on it, or partition it somehow... You'd need 8 servers. Not sure if this is something I'd recommend... (all on one server even if it was possible). I assume the instances are fairly small.... but if you loose one server you'd loose all your replicas.... Not sure what you're trying to achieve, but having a set of servers spread across different locations with HDR/RSS could be better... Regards. On Wed, Oct 19, 2016 at 5:59 PM, Kate Tomchik <katedarts@gmail.com> wrote: > Hey all, > I am looking into the feasibility of using HDR from 2000 sites to a central > server. Is it possible for a single server to hold 2000 instances, and has > anyone else out there done that? Any details you have on how you run > queries against that as well would be great info. > > -- > Will chat again soon, > Kate > katedarts@gmail.com > > --001a1144491cc9cb96053f3ab9e0 > > > ************************************************************ > ******************* > Forum Note: Use "Reply" to post a response in the discussion forum. > > -- Fernando Nunes Portugal http://informix-technology.blogspot.com My email works... but I don't check it frequently... --001a114239e6073404053f3ad2c3

I think you=E2=80=99re missing something kind of important. Each HDR = instance as to be an exact replica of the primary, including disk paths. = You may be using sym links (I hope), but that means that on a server, = only one instance can be using /opt/IBM/disk/my_sym_link to point to a = storage device wherever it might be. If each HDR instance is using the = same disk paths, each instance MUST be on a separate server whether = physical or VM or LPAR. So the question becomes, can a VM server = support 2k separate VMs? Beyond that, to query a large group of instances requires using Grid = Queries which requires putting the primaries into a Flex Grid cluster. Carlton ------------------------------------------------- Carlton Doe Flower Mound, TX No trees were killed in the transmission of this email, however a large = number of electrons were temporarily inconvenienced.=20 > On Oct 19, 2016, at 11:59 AM, Kate Tomchik <katedarts@gmail.com> = wrote: >=20 > Hey all,=20 > I am looking into the feasibility of using HDR from 2000 sites to a = central=20 > server. Is it possible for a single server to hold 2000 instances, and = has=20 > anyone else out there done that? Any details you have on how you run=20= > queries against that as well would be great info.=20 >=20 > --=20 > Will chat again soon,=20 > Kate=20 > katedarts@gmail.com=20 >=20 > --001a1144491cc9cb96053f3ab9e0=20 >=20 >=20 > = **************************************************************************= *****=20 > Forum Note: Use "Reply" to post a response in the discussion forum.=20= >=20

Very good point, we do use links though so we could do some renaming. In general it seems to be harder solution than ER, we just have to wait for version 12 to try that as our version has a bug with unique index ER. On Wed, Oct 19, 2016 at 1:13 PM, Carlton Doe <dbaresrc@xmission.com> wrote: > I think you=E2=80=99re missing something kind of important. Each HDR = > instance as to be an exact replica of the primary, including disk paths. = > You may be using sym links (I hope), but that means that on a server, = > only one instance can be using /opt/IBM/disk/my_sym_link to point to a = > storage device wherever it might be. If each HDR instance is using the = > same disk paths, each instance MUST be on a separate server whether = > physical or VM or LPAR. So the question becomes, can a VM server = > support 2k separate VMs? > > Beyond that, to query a large group of instances requires using Grid = > Queries which requires putting the primaries into a Flex Grid cluster. > > Carlton > > ------------------------------------------------- > Carlton Doe > Flower Mound, TX > > No trees were killed in the transmission of this email, however a large = > number of electrons were temporarily inconvenienced.=20 > > > On Oct 19, 2016, at 11:59 AM, Kate Tomchik <katedarts@gmail.com> = > wrote: > >=20 > > Hey all,=20 > > I am looking into the feasibility of using HDR from 2000 sites to a = > central=20 > > server. Is it possible for a single server to hold 2000 instances, and = > has=20 > > anyone else out there done that? Any details you have on how you run=20= > > > queries against that as well would be great info.=20 > >=20 > > --=20 > > Will chat again soon,=20 > > Kate=20 > > katedarts@gmail.com=20 > >=20 > > --001a1144491cc9cb96053f3ab9e0=20 > >=20 > >=20 > > = > ************************************************************ > **************= > *****=20 > > Forum Note: Use "Reply" to post a response in the discussion forum.=20= > > >=20 > > > ************************************************************ > ******************* > Forum Note: Use "Reply" to post a response in the discussion forum. > > -- Will chat again soon, Kate katedarts@gmail.com --bcaec54eeb56b47a2b053f3b20e1

Note that symbolic links usually are used to allow the change of the underlying raw or cooked devices. They don't allow you to change the PATH used in "onspaces" or equivalent. With 2000 instances I assume a greta degree of standardization was used, so I wouldn't be surprised if the 2000 instances use the same PATHs for the chunks (links). If that was the case each instance would have to be changed.... for example include the server name or mahcine name or geographical location, or store number or whatever distinguishes them in the PATH. A way to do that would be to simulate an external backup/restore withe rename of the chunks.... That would be quick and easy... and no real backup/restore would have to be done. Regards. On Wed, Oct 19, 2016 at 6:27 PM, Kate Tomchik <katedarts@gmail.com> wrote: > Very good point, we do use links though so we could do some renaming. In > general it seems to be harder solution than ER, we just have to wait for > version 12 to try that as our version has a bug with unique index ER. > > On Wed, Oct 19, 2016 at 1:13 PM, Carlton Doe <dbaresrc@xmission.com> > wrote: > > > I think you=E2=80=99re missing something kind of important. Each HDR = > > instance as to be an exact replica of the primary, including disk paths. > = > > You may be using sym links (I hope), but that means that on a server, = > > only one instance can be using /opt/IBM/disk/my_sym_link to point to a = > > storage device wherever it might be. If each HDR instance is using the = > > same disk paths, each instance MUST be on a separate server whether = > > physical or VM or LPAR. So the question becomes, can a VM server = > > support 2k separate VMs? > > > > Beyond that, to query a large group of instances requires using Grid = > > Queries which requires putting the primaries into a Flex Grid cluster. > > > > Carlton > > > > ------------------------------------------------- > > Carlton Doe > > Flower Mound, TX > > > > No trees were killed in the transmission of this email, however a large = > > number of electrons were temporarily inconvenienced.=20 > > > > > On Oct 19, 2016, at 11:59 AM, Kate Tomchik <katedarts@gmail.com> = > > wrote: > > >=20 > > > Hey all,=20 > > > I am looking into the feasibility of using HDR from 2000 sites to a = > > central=20 > > > server. Is it possible for a single server to hold 2000 instances, and > = > > has=20 > > > anyone else out there done that? Any details you have on how you > run=20= > > > > > queries against that as well would be great info.=20 > > >=20 > > > --=20 > > > Will chat again soon,=20 > > > Kate=20 > > > katedarts@gmail.com=20 > > >=20 > > > --001a1144491cc9cb96053f3ab9e0=20 > > >=20 > > >=20 > > > = > > ************************************************************ > > **************= > > *****=20 > > > Forum Note: Use "Reply" to post a response in the discussion forum.=20= > > > > >=20 > > > > > > ************************************************************ > > ******************* > > Forum Note: Use "Reply" to post a response in the discussion forum. > > > > > > -- > Will chat again soon, > Kate > katedarts@gmail.com > > --bcaec54eeb56b47a2b053f3b20e1 > > > ************************************************************ > ******************* > Forum Note: Use "Reply" to post a response in the discussion forum. > > -- Fernando Nunes Portugal http://informix-technology.blogspot.com My email works... but I don't check it frequently... --001a113f97722eb30a053f3b4d59

Hi Kate You cannot have more than 256 instances on a server because of the Dbservernum that is used to create the key for the shared memory segments. The servernum is in a single byte ( FF in hex is the maximum that you can put in a byte). You can may be partition your server and install several lpars on IBM or VMware virtualizations and something similar . That way you would have a single big server and each partition will have its own OS that can host 256 instances per partition. Hope that this helps Khaled Bentebal de mon portable > Le 19 oct. 2016 à 18:59, Kate Tomchik <katedarts@gmail.com> a écrit : > > Hey all, > I am looking into the feasibility of using HDR from 2000 sites to a central > server. Is it possible for a single server to hold 2000 instances, and has > anyone else out there done that? Any details you have on how you run > queries against that as well would be great info. > > -- > Will chat again soon, > Kate > katedarts@gmail.com > > --001a1144491cc9cb96053f3ab9e0 > > > ******************************************************************************* > Forum Note: Use "Reply" to post a response in the discussion forum. >

blockquote, div.yahoo_quoted { margin-left: 0 !important; border-left:1px #715FFA solid !important; padding-left:1ex !important; background-color:white !important; } You can have only 255 instances on a given logical server. That's because the server ID is used to calculate the shared memory key and the server ID is only contained in one byte of that key. If you want to have more instances on a physical machine, then you would need to create virtual machines al la VMWARE. Also, as Carlton mentioned you must use the same path name for the chunks on the primary as the secondary. That would be an issue unless you are using relative path names for the chunks. But let's focus on what the goal is. Is this really for high availability, or are you trying to consolidate the data from the remote machines? If the real goal is data consolidation, then I would suggest that you consider using parallel sharing that exists in IDS 12. The advantage of using that technique is that there does not have to be any data movement at all and to perform a query against all 2000 servers would not take much longer than a query against an individual server since all aspects of the query would be performed in parallel. In late versions of IDS 11, we introduced the grid query which is basically the first phase of the parallel shard query. The biggest difference is that the grid query requires dynamically moving the remote data dictionary to the server on which the query is executed and the grid query performs the operation on each node serially. With the shard query none of that happens. There is a form of the shard table definition called "informational". This just updates the shard dictionary but does NOT cause the data to be moved from one node to the other. Also - while the parallel shard query is described with the nosql features, it is NOT restricted to nosql tables. Sent from Yahoo Mail for iPad On Wednesday, October 19, 2016, 11:59 AM, Kate Tomchik <katedarts@gmail.com> wrote: Hey all, I am looking into the feasibility of using HDR from 2000 sites to a central server. Is it possible for a single server to hold 2000 instances, and has anyone else out there done that? Any details you have on how you run queries against that as well would be great info. -- Will chat again soon, Kate katedarts@gmail.com --001a1144491cc9cb96053f3ab9e0 ******************************************************************************* Forum Note: Use "Reply" to post a response in the discussion forum.