HDR question once again

Hi All, I have a Solaris 7 + IDS 9.4UC6 as primary and a secondary server (Solaris 7 + IDS 9.4UC6. It is running smoothly and fine. However, my boss inquire from me if it's possible to have the following setup: 1. Primary server (IDS 9.4UC6 + Solaris 7) Secondary Server (IDS 11.x + Solaris 10.x) Would it be possible primary and secondary have different versions? 2. I've already asked this but, Is it possible to have a 2nd secondary server using version 9.4UC6? The 2nd Secondary server will have the same setup as the existing secondary server. There will be 2 secondary servers connected to the primary server. 3. Is it possible to have another primary server or secondary server that will serve as standby database? This standby database will be updated once per week in order to sync the database. Thank you in advance.

See my comments in line below: However, it would be a VERY good idea to consider relicensing and upgrading to 11.50 or 11.70. There are currently special licensing deals for upgrades that include support on the older out-of-support versions during a defined upgrade period. You could use the ER scenario described in your first question to create the new server then swap primary and secondary to bring the new version online. Art Art S. Kagel Advanced DataTools (www.advancedatatools.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 my employer, Advanced DataTools, 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 Thu, Feb 9, 2012 at 3:39 AM, JACK PAPA <informix2009@gmail.com> wrote: > Hi All, > > I have a Solaris 7 + IDS 9.4UC6 as primary and a secondary server (Solaris > 7 + > IDS 9.4UC6. It is running smoothly and fine. However, my boss inquire from > me > if it's possible to have the following setup: > > 1. > Primary server (IDS 9.4UC6 + Solaris 7) > Secondary Server (IDS 11.x + Solaris 10.x) > > Would it be possible primary and secondary have different versions? > Using Enterprise Replication the versions can be different and 11.xx does support 9.40 replicants, yes. Either could be primary or even peers, however, you will have to setup the replication using the cdr command line utility from 11.xx. > > 2. I've already asked this but, Is it possible to have a 2nd secondary > server > using version 9.4UC6? The 2nd Secondary server will have the same setup as > the > existing secondary server. There will be 2 secondary servers connected to > the > primary server. > Using ER it is possible, but not using HDR. The 9.40 versions did not support RSS secondaries which is what you would need to create the additional secondary servers. > > 3. Is it possible to have another primary server or secondary server that > will > serve as standby database? This standby database will be updated once per > week > in order to sync the database. > You can always take an archive on the primary are restore it once a week to create a standby server, yes. In the current releases (so 11.50 & 11.70) you can use a new feature called Continuous Log Restore to maintain the standby secondary at near real-time readiness by making an initial restore then restoring the logical log archives as they are made from the primary. > > Thank you in advance. > > > > ******************************************************************************* > Forum Note: Use "Reply" to post a response in the discussion forum. > > --00151747913a1e586f04b886274c

Hi Art, Let me first thank you for all of your answers. I think you have answered my questions and what i want to achieve is to have a (2) two secondary servers connected to 1 primary server. I am a little bit confused on what is the difference between an Enterprise Replication and High Availability Data Replication. So far, what i have seen is 1 primary server and (1) Secondary Server connected which performs log shipping. The primary server is in Read Write mode, while the Secondary server is in Read Only mode. I've never used "cdr" command before, instead i always used "onmode -d" command. I am reading Enterprise Replication Guide v. 9.4 to understand more of what i need, but still can't differentiate ER and HDR. Anyways, I'll do a follow up post on this of what i read. Regards,

Jack: You are missing a lot. A bit of recent history then. Informix from version 7.10 through 7.31 and 9.10 through version 10.00, supported two overall replication types: 1. HDR or High availability Data Replication - This supported one read/write primary server and one read-only HDR secondary server. In this scenario the entire server (or at least any logged databases in the server) were replicated by making an initial copy of the primary using ontape or onbar and using onmode -d to tell each server about the other. After that the primary sent copies of each logical log buffer to the secondary before writing it to disk locally. The replication could be synchronous (transactions on the primary wait for the secondary to acknowledge that the log record containing its COMMIT had been applied on the secondary before applying the COMMIT on the primary) or asynchronous (the primary does not wait to commit the local transaction). HDR can only support servers which are all on the same platform and running the same exact version of the Informix software. 2. Enterprise Replication or ER. This replicates data at the table level (and even at the row level as filters can determine which rows to replicate) in many different configurations of from two to hundreds of replicated servers. Replication can be master-to-slave, peer-to-peer and can replicate all of the data or only part of the data in one or all tables. It is more complex to set up, but far more flexible and does not require an archive and restore to initiate the replication (an new server can start out empty and ER will provision it with data). All server are read-write and data can be distributed to many nodes from a single root source, gathered from many nodes to a central target, or replicated in a complex network of peer servers. ER supported servers on multiple platforms running different versions of the Informix software (within reason - the oldest versions couldn't keep up with more recent ones, but IB that from 7.30 & 9.21 on cross version replication was possible). ER nodes and replicated tables are configured using a utility called cdr. (You have it. If you run it cdr with no args will print out the commands that it supports.) 3. Combinations of ER and HDR can be configured where any ER node can actually be an HDR pair. With the release of Informix 11.10 IBM introduced two new kinds of secondary servers and renamed the whole package "MACH11" an acronym which I refuse to expand. ;-) The new server types were: 1. RSS or Remote Secondary Server - This is identical to an HDR secondary except that one can still have only one HDR secondary but can have zero to many RSS secondaries. While the communications with HDR secondaries was and remains half duplex (each packet has to be acked before another is sent) RSS secondaries operate under a full duplex protocol permitting more efficient operations over slower and less reliable communications. (An RSS secondary is more likely to work in a remote data center at the other end of the world than an HDR secondary.) Logical log buffers are shipped to the RSS secondaries just as if they were HDR secondaries, but, the primary does not wait for any acks and will spool the buffers locally and reship them later if the RSS secondaries fall behind. RSS secondaries can only be promoted to an HDR secondary, not directly to a primary. 2. SDS or Shared Disk Secondary Server - These secondaries share the disks used by the primary. They can be instances running on the same physical host connecting locally to the disk chunks, running on different hosts connecting to the chunks on a SAN or NAS that is sharable, or running in a remote data center connected to chunk files that are replicated at the SAN level between data centers. When the primary modifies a page on disk it send ONLY the page's address to all SDS secondaries that it knows about which then reread the page into their own private cache if they had a stale version of the page in cache at the time. No data moves to the SDS secondaries. SDS secondaries have private TEMP dbspaces for their own sorting and temp tables which can be on local disk. Any SDS secondary can replace a down primary and be promoted to primary over the remaining SDS, RSS, and HDR secondaries. (You can still have the HDR secondary take over by converting an SDS secondary to primary and then switching roles with the HDR secondary). Since there is no lock sharing, unlike Oracle RAC clusters, SDS secondaries do not require a high speed backplane to work efficiently and SDS scales to well over a hundred secondaries in the lab and I have heard of dozens of secondaries to a single primary being configured in the field. RSS secondaries are initialized just like HDR secondaries. SDS secondaries just require the Informix software, an appropriately configured ONCONFIG file, and shared access to the primary's chunk files. They are enabled using onmode -d sub-commands. ER was improved with the concept of templates which let you more easily replicate many tables at once. With the release of Informix 11.50 RSS and SDS secondary servers became writable! Now, if you want, users can connect to any secondary server, whether it is an HDR, RSS, or SDS secondary and modify the data in their tables. The secondary sends the update to the primary which performs the work and updates the secondaries appropriately. Once the primary acknowledges the transaction the secondary will notify the user. Version 11.50 also introduced the ability to create a new secondary server as a clone of an existing server in a single step with the new ifxclone utility. Informix version 11.70 introduced the concept of Informix Flexible Grid which improved the interoperability of ER, HDR, RSS, and SDS so that you can manage an entire grid of heterogeneous primary, secondary, and peer servers using new 'grid' commands added to the cdr utility. You can manage the grid as a whole including CREATEing, DROPping, and ALTERing tables, creating indexes, etc. across every server in the grid with a single command. Flexible grid also allows you to perform rolling upgrades of the servers in the grid with no down time. Finally, 11.70 introduced the concept of Transaction Survival. Transactions started by a client connected to a secondary server (so HDR, RSS, or SDS) continue uninterrupted if the primary server crashes. The transactions are picked up by whichever secondary server takes over as primary (HDR or SDS) and continue after a brief pause as if nothing happened. Art Art S. Kagel Advanced DataTools (www.advancedatatools.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 my employer, Advanced DataTools, 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 ind

Hi Art, Thank you for your explanation of Informix technology. I really appreciate it. I just can't imagine, i wasn't able to explore this Informix functionalities even way back v. 7.3. Most of my setup is just an OLTP, single server with no replication. Getting back to your explanation, it shows th at we have an HDR setup as of this moment. I've checked primary and secondary server via "cdr list server" and i didn't see any configuration. using "onstat -g dri", shows the status and where it is connected. Given this scenario above that we have HDR which uses version 9.4UC6 (and not ER),per your explanation, does it mean, we CAN'T have a "2nd second secondary server connected to primary server"? Can we have a new ER (table level), connected to the primary server? or connected to existing secondary server? With regards to v 11.5/11.7, i have already recommended it to the company but I was asked to find another solution to the issue. Kudos to you Art. Great! Thank you,

Your only solutions would be to make your primary an ER root server and add another ER node (or a second primary-HDR pair replicated by ER from the current one), or you could use a third party replication product to create a secondary. Hit Software's DbMoto works with Informix, but since 9.40 did not support the new Change Data Capture API it works by adding table level triggers into your databases. You may need to get an older version, the latest release uses CDC for Informix v11.xx and I don't know if it still supports the trigger method. Either way the replication would be working off of the HDR primary (the secondary doesn't fire triggers for DbMoto and the HDR secondary knows it is a secondary so it won't send any data to an ER node until it is promoted to primary). As far as I know, those are the only ways to add more secondary servers. If you decide to go ER make sure you are reading the 9.40 documentation since there is lots of new stuff in later ER releases that your system won't support. Don't want to keep butting your head against walls. ;-) Art Art S. Kagel Advanced DataTools (www.advancedatatools.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 my employer, Advanced DataTools, 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 Fri, Feb 10, 2012 at 4:38 AM, JACK PAPA <informix2009@gmail.com> wrote: > Hi Art, > > Thank you for your explanation of Informix technology. I really appreciate > it. > I just can't imagine, i wasn't able to explore this Informix > functionalities > even way back v. 7.3. Most of my setup is just an OLTP, single server with > no > replication. > > Getting back to your explanation, it shows th at we have an HDR setup as of > this moment. I've checked primary and secondary server via "cdr list > server" > and i didn't see any configuration. using "onstat -g dri", shows the status > and where it is connected. > > Given this scenario above that we have HDR which uses version 9.4UC6 (and > not > ER),per your explanation, does it mean, we CAN'T have a "2nd second > secondary > server connected to primary server"? > > Can we have a new ER (table level), connected to the primary server? or > connected to existing secondary server? > > With regards to v 11.5/11.7, i have already recommended it to the company > but > I was asked to find another solution to the issue. > > Kudos to you Art. Great! > > Thank you, > > > > ******************************************************************************* > Forum Note: Use "Reply" to post a response in the discussion forum. > > --e89a8f22c681c847ea04b89a9e36

Thanks Art. I have read the ER setup and it looks tedious to setup specially if you have an HDR in place (running for several years). I don't want to mess up the whole system. It's like i'm trying to put a solution to a problem and later on the solution will become a problem.

Understood. Upgrading so you can take advantage of RSS and SDS secondaries is the simpler solution. I understand your bosses reluctance to relicense Informix. It will not be cheap, but there are advantages including up to 40% better performance on the same hardware and lots of new features some geared for additional performance gains where they can be deployed. Good luck and let me know if I can help. Ping me privately if you need help convincing Mgt. to upgrade. Art Art S. Kagel Advanced DataTools (www.advancedatatools.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 my employer, Advanced DataTools, 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 Tue, Feb 14, 2012 at 8:05 AM, JACK PAPA <informix2009@gmail.com> wrote: > Thanks Art. > > I have read the ER setup and it looks tedious to setup specially if you > have > an HDR in place (running for several years). I don't want to mess up the > whole > system. It's like i'm trying to put a solution to a problem and later on > the > solution will become a problem. > > > > ******************************************************************************* > Forum Note: Use "Reply" to post a response in the discussion forum. > > --e89a8f3ba4290adc7804b8ed2d7d