UK IUG Meeting

any news from today's uk iug meeting, which i was unfortunately unable to make? -- Neil Truby t:01932 724027 Director m:07798 811708 Ardenta Limited e:neil.truby@ardenta.com

I missed the start due to traffic - only allowed 2.5 hours to get there, silly me! I caught the end of the talk by the IBM senior person (senior North Europe Sales I think) - did anyone get her name? The end of her talk mentioned - DB/s and IDS Development Labs are seperate but can share ideas. - The hot thing is "Master Data". We can read from lots of data repositories but we also need to be able to keep the duplicate data across the repositories in sync in realtime! There was another "name change" but I missed what it was. Carlton Doe replaced Mark Scranton due to a family emergency. Carlton did a good job of going through the important new stuff in IDS 10 but did skip over the lesser bits as I think he had too many slides from Mark! Carlton definitely listened to what people wanted and tried to cover the bits that are important to the people in the room and certainly kept our interest! Madison did a great presentation on replication including a live demo on how to setup ER for a database with just a few commands and how to add a node to replication with just a few commands. ER can now create databases/tables on the other nodes for you! Also in IDS 10 ER has commands to a) verify the ER setup across all nodes in one go rather than starting ER and waiting for stuff to fail! b) detect inconsistent rows across all nodes and produce a report of missing/different rows on each node and c) repair inconsistent rows on all nodes! Finally the Q&A session at the end answered everything that anyone raised. One question on OLDB support was deferred for a more detailed answer but the person who took it was high up in UK support (email me mate I didn't get the name!) so except an answer soon! Apparently OLDB needs to be installed in the IDS server but then the question is what happens if the clients are running different versions than the server. Watch this space for more info... The main questions from the floor included - Connectivity questions - OLDB, .NET what is being done about improving these. - How do people know when fixpack (UC1 UC2 UC3 etc) releases come out - answer check IBM's website for the Product Availablity Spreadsheet and Release Notes/Evaluation copies. Andy Legge mentioned the spreadsheet has been updated something like 8 times this year so it is being actively maintained now! Two cool IDS 10 features I missed before are a) archecker and table level restore can be used to restore from archives on a different platform! You could restore a complete database from an archive on a different platform! b) If you setup groupnames in your sqlhosts file you can connect to the groupname. In an HDR setup this will always connect to whatever node is the primary. It talks to both nodes and finds which one is currently the primary. One HDR weakness is the "split brain " syndrome where the two nodes can no longer talk to one another and both become primary but they can both talk to the clients. Perhaps this should be resolved outside the IDS server by an arbitrator(?) so potentially look for more work in this area in future versions!! The slides should be getting sent to all committee members soon and we'll see what we do about making them available.

Hi, I have a question for you regarding High Availability Data Replication, a feature found in DB2. How does Informix's feature set of replication and HA compare with DB2 HADR? We are currently evaluating DB2 HADR and were curious if Informix had different/better/other features worth investigating. We like what DB2 has, is Informix better? Thanks in advance! PS DB2 also has the redirected-restore feature, where you can restore a database on to a different machine from a backup. david@smooth1.co.uk wrote: > I missed the start due to traffic - only allowed 2.5 hours to get > there, silly me! > > I caught the end of the talk by the IBM senior person (senior North > Europe Sales I think) - did anyone get her name? > > The end of her talk mentioned > > - DB/s and IDS Development Labs are seperate but can share ideas. > - The hot thing is "Master Data". We can read from lots of data > repositories but we also need to be able to keep the duplicate data > across the repositories in sync in realtime! > > There was another "name change" but I missed what it was. > > Carlton Doe replaced Mark Scranton due to a family emergency. > > Carlton did a good job of going through the important new stuff in IDS > 10 but did skip over the lesser bits as I think he had too > many slides from Mark! Carlton definitely listened to what people > wanted and tried to cover the bits that are important to the > people in the room and certainly kept our interest! > > Madison did a great presentation on replication including a live demo > on how to setup ER for a database with just a few commands > and how to add a node to replication with just a few commands. ER can > now create databases/tables on the other nodes for you! > > Also in IDS 10 ER has commands to > > a) verify the ER setup across all nodes in one go rather than starting > ER and waiting for stuff to fail! > > b) detect inconsistent rows across all nodes and produce a report of > missing/different rows on each node and > > c) repair inconsistent rows on all nodes! > > Finally the Q&A session at the end answered everything that anyone > raised. > > One question on OLDB support was deferred for a more detailed answer > but the person who took it was high up in UK support > (email me mate I didn't get the name!) so except an answer soon! > Apparently OLDB needs to be installed in the IDS server but > then the question is what happens if the clients are running different > versions than the server. Watch this space for more info... > > The main questions from the floor included > > - Connectivity questions - OLDB, .NET what is being done about > improving these. > - How do people know when fixpack (UC1 UC2 UC3 etc) releases come out - > answer check IBM's website for the Product > Availablity Spreadsheet and Release Notes/Evaluation copies. Andy Legge > mentioned the spreadsheet has been updated > something like 8 times this year so it is being actively maintained > now! > > Two cool IDS 10 features I missed before are > > a) archecker and table level restore can be used to restore from > archives on a different platform! You could restore a complete > database from an archive on a different platform! > > b) If you setup groupnames in your sqlhosts file you can connect to the > groupname. In an HDR setup this will always connect to > whatever node is the primary. It talks to both nodes and finds which > one is currently the primary. One HDR weakness is the "split > brain " syndrome where the two nodes can no longer talk to one another > and both become primary but they can both talk to the > clients. Perhaps this should be resolved outside the IDS server by an > arbitrator(?) so potentially > look for more work in this area in future versions!! > > The slides should be getting sent to all committee members soon and > we'll see what we do about making them available. >

<david@smooth1.co.uk> wrote in message news:1131585243.973442.263950@g47g2000cwa.googlegroups.com... > I caught the end of the talk by the IBM senior person (senior North > Europe Sales I think) - did anyone get her name? Michelle Unger, if I remember correctly. > There was another "name change" but I missed what it was. The DB2 division/brand (including Informix) has been renamed "Information Management"! -- Regards, Doug Lawry www.douglawry.webhop.org

Informix has had HADR since 1994 I guess DB2s is pretty similar since it was based on the Informix design. i wouldn't like to say that either is "better". It should be fair to say Informix's is more mature. Informix can also do restores onto a different machine from a backup. if you mean while the restore is being performed as the backup is being performed that should be achievable using unix pipes.

Thanks for this information, and to the rest of you who posted. It was interesting to note that Karl mentioned Informix HADR can read the standby database. Can you perform backups? ( I know I need to read the docs ) We were told we can split-mirror the stand-by on DB2 to do backups but don't want to get into that for a variety of reasons at this time. I have noticed that DB2's HADR is very robust, and easy to set up and manage. The next step is for me to get some kind of RedBook on Informix's HADR. Thanks to all of you! scottishpoet wrote: > Informix has had HADR since 1994 > > I guess DB2s is pretty similar since it was based on the Informix > design. i wouldn't like to say that either is "better". It should be > fair to say Informix's is more mature. > > Informix can also do restores onto a different machine from a backup. > > if you mean while the restore is being performed as the backup is being > performed that should be achievable using unix pipes. >

RollForward Wizard wrote: > Thanks for this information, and to the rest of you who posted. > It was interesting to note that Karl mentioned Informix HADR can > read the standby database. Can you perform backups? ( I know I > need to read the docs ) We were told we can split-mirror the > stand-by on DB2 to do backups but don't want to get into that > for a variety of reasons at this time. > > I have noticed that DB2's HADR is very robust, and easy to set > up and manage. The next step is for me to get some kind of > RedBook on Informix's HADR. > > Thanks to all of you! > Interesting question : > Can you perform backups? Presumably you means backups on the secondary. No is the answer; but I think it more relevant to ask "Why do you want to do backups on the secondary?" (i.e. the "backup" server in informix terms). Is it because of the way DB2 does backups? Is it because you want to take the load of running a backup off the "primary" (informix parlance)? And, I wonder about the "perpetual rollforward" thing as well, which informix is not documented to be able to do (but with a little inventiveness can be done). Still ... Just a potato

TBP wrote: > RollForward Wizard wrote: >> Thanks for this information, and to the rest of you who posted. >> It was interesting to note that Karl mentioned Informix HADR can >> read the standby database. Can you perform backups? ( I know I >> need to read the docs ) We were told we can split-mirror the >> stand-by on DB2 to do backups but don't want to get into that >> for a variety of reasons at this time. >> >> I have noticed that DB2's HADR is very robust, and easy to set >> up and manage. The next step is for me to get some kind of >> RedBook on Informix's HADR. >> >> Thanks to all of you! >> > Interesting question : > > > Can you perform backups? > > Presumably you means backups on the secondary. > > No is the answer; but I think it more relevant to ask "Why do you want > to do backups on the secondary?" (i.e. the "backup" server in informix > terms). > Good question. My client wants to set up split-mirror off of the secondary to do "backup snapshots" per DB2/SAP guidance. We're telling them to stay away from split mirror frankly because they don't have the peoples to manage split mirror, and it's yet another point of failure they can't manage even if they wanted to. They thought by backing up off the secondary it would take the load off of the primary server. ( Note: this is an SAP-on-Windows server ) I've indicated to them that HADR is essentially a live/hot backup anyway, so split mirror is a bit luxurious/redundant. > Is it because of the way DB2 does backups? DB2 doesn't allow backing up off the secondary, but the guidance says it can be done via split-mirroring the secondary. This is great if you have someone around to manage the SAN, they don't have anyone on site to manage the SAN. > Is it because you want to take the load of running a backup off the > "primary" (informix parlance)? > Yes, see above. > And, I wonder about the "perpetual rollforward" thing as well, which > informix is not documented to be able to do (but with a little > inventiveness can be done). > We like the "chimp" factor for DB2 HADR, a chimp could manage it if need be, just install the Tivoli Banana Services Management Console, and it's just beautiful. :-) ( Has to be installed after SAP to get support ) > Still ... > > Just a potato

"TBP" <TheBigPotato@NotHere.Co.Uk> wrote in message news:T51df.1144$fN5.291@newsfe6-gui.ntli.net... > RollForward Wizard wrote: >> Thanks for this information, and to the rest of you who posted. >> It was interesting to note that Karl mentioned Informix HADR can >> read the standby database. Can you perform backups? ( I know I >> need to read the docs ) We were told we can split-mirror the >> stand-by on DB2 to do backups but don't want to get into that >> for a variety of reasons at this time. >> >> I have noticed that DB2's HADR is very robust, and easy to set >> up and manage. The next step is for me to get some kind of >> RedBook on Informix's HADR. >> >> Thanks to all of you! >> > Interesting question : > > > Can you perform backups? > > Presumably you means backups on the secondary. > > No is the answer; but I think it more relevant to ask "Why do you want > to do backups on the secondary?" (i.e. the "backup" server in informix > terms). > > Is it because of the way DB2 does backups? > Is it because you want to take the load of running a backup off the > "primary" (informix parlance)? Because the replicate is on a stand-by box on a different subnet, and that box also houses multiple development environments that ytou would like to refresh frequently from live without the bandwidth and delay problems of sending a full archive down the WAN?

Neil Truby wrote: > "TBP" <TheBigPotato@NotHere.Co.Uk> wrote in message >> Is it because of the way DB2 does backups? >> Is it because you want to take the load of running a backup off the >> "primary" (informix parlance)? > > > Because the replicate is on a stand-by box on a different subnet, and that > box also houses multiple development environments that ytou would like to > refresh frequently from live without the bandwidth and delay problems of > sending a full archive down the WAN? > > Wouldn't you just use standard replication for this kind of feature? Search IBM's RedBooks for sg246828.pdf ( HADR is not the same replication that I think you are suggesting. )