1824623
Members
4079
Online
109672
Solutions
"...from way back since 1084."
Wow!
Hope this helps!
Regards
Torsten.
__________________________________________________
There are only 10 types of people in the world -
those who understand binary, and those who don't.
__________________________________________________
No support by private messages. Please ask the forum!
If you feel this was helpful please click the KUDOS! thumb below!
Wow!
Hope this helps!
Regards
Torsten.
__________________________________________________
There are only 10 types of people in the world -
those who understand binary, and those who don't.
__________________________________________________
No support by private messages. Please ask the forum!
If you feel this was helpful please click the KUDOS! thumb below!
Nelson,
>> Do you agree that there really is NO True High Availability (No APP DownTIme) solution for HP-UX 11.11 to 11.31 ?
* By your definition of "True High Availability" no there isn't - but then neither is there for any other flavour of UNIX.
>> What happened to the Comapq Purchase and the Tru64 (Tru-CLuster) crown jewel and promise of a zero Applications Downtime UNIX from HP?
* There were 2 features of TruCluster that were to make their way into Serviceguard - a Cluster Filesystem and Single System Image (SSI). The Cluster Filesystem was delivered instead by using the Veritas CFS solution intergrated with Serviceguard. SSI was always a 'manageability feature' rather than a specific HA feature, and has largely been obsoleted by advancements in managment tools such as SIM... although you can pick specific areas where a single system image would still be preferrable.
>> But isn't one other leading UNIX claim to have a Live Partition Mobility a step closer to having Applications basically "LIVE" in a fluid server pool -- a realisation of TRUE Availability? Although NEW - this is promising and is catching my clients attention.
* Read between the lines and you'll find that on AIX you can have Live Partition Mobility *or* HACMP, but not both... so you can either protect yourself from planned downtime or unplanned downtime but not both at the same time... doesn't sound very HA to me. Also IBM needed live partition mobility because every time you need to do HW manintenance on a p-series system you have to drop the entire box as there's nor electrical isolation between lpars. On HP-UX we have this feature and can take nPars offline for maintenance without dropping an entire complex.
> During HPTF 2007, there was mention that the 11.31 only version of ServiceGuard will have vMotion like features... Man if that will be true - HP-UX may have a fight in the UNIX battles ahead.
* It's coming at some point - speak to your HP rep and maybe you can get an NDA demo of these kind of features - but don't expect to be able to sell to your customers this year.
HTH
Duncan
I am an HPE Employee
>> Do you agree that there really is NO True High Availability (No APP DownTIme) solution for HP-UX 11.11 to 11.31 ?
* By your definition of "True High Availability" no there isn't - but then neither is there for any other flavour of UNIX.
>> What happened to the Comapq Purchase and the Tru64 (Tru-CLuster) crown jewel and promise of a zero Applications Downtime UNIX from HP?
* There were 2 features of TruCluster that were to make their way into Serviceguard - a Cluster Filesystem and Single System Image (SSI). The Cluster Filesystem was delivered instead by using the Veritas CFS solution intergrated with Serviceguard. SSI was always a 'manageability feature' rather than a specific HA feature, and has largely been obsoleted by advancements in managment tools such as SIM... although you can pick specific areas where a single system image would still be preferrable.
>> But isn't one other leading UNIX claim to have a Live Partition Mobility a step closer to having Applications basically "LIVE" in a fluid server pool -- a realisation of TRUE Availability? Although NEW - this is promising and is catching my clients attention.
* Read between the lines and you'll find that on AIX you can have Live Partition Mobility *or* HACMP, but not both... so you can either protect yourself from planned downtime or unplanned downtime but not both at the same time... doesn't sound very HA to me. Also IBM needed live partition mobility because every time you need to do HW manintenance on a p-series system you have to drop the entire box as there's nor electrical isolation between lpars. On HP-UX we have this feature and can take nPars offline for maintenance without dropping an entire complex.
> During HPTF 2007, there was mention that the 11.31 only version of ServiceGuard will have vMotion like features... Man if that will be true - HP-UX may have a fight in the UNIX battles ahead.
* It's coming at some point - speak to your HP rep and maybe you can get an NDA demo of these kind of features - but don't expect to be able to sell to your customers this year.
HTH
Duncan
I am an HPE Employee
>> 2.) As far as Database High Availability - is SGeRAC better than just 10G RAC?
Bear in mind my response is subjective as an HP'ite - I'd suggest the following 'advantages' to SGeRAC over straight RAC:
i) You can run on a cluster filessystem and using ODM still obtain performance within 5% of raw.
ii) You don't have to use ASM (I consider this a *good* thing at the moment - ASM is a pretty new product compared to LVM and VxVM - no doubt it will improve, but right now its early days... no multipathing and no integration with array based snapshot technologies being the obvious big issues)
iii) Serviceguard benefits from kernel integration - this makes a real difference during hung node detection - a big deal in clusters.
iv) Serviceguard has built in HA networking and in the latest versions can also monitor and respond to issues with RAC's cluster interconnect.
v) And of course HA doesn't begin and end with the database - if you want to cluster application components Serviceguard is a tried and trusted solution - I'm yet to come across anyone who has used Oracle's Clusterware for anything other than RAC.
HTH
Duncan
I am an HPE Employee
Bear in mind my response is subjective as an HP'ite - I'd suggest the following 'advantages' to SGeRAC over straight RAC:
i) You can run on a cluster filessystem and using ODM still obtain performance within 5% of raw.
ii) You don't have to use ASM (I consider this a *good* thing at the moment - ASM is a pretty new product compared to LVM and VxVM - no doubt it will improve, but right now its early days... no multipathing and no integration with array based snapshot technologies being the obvious big issues)
iii) Serviceguard benefits from kernel integration - this makes a real difference during hung node detection - a big deal in clusters.
iv) Serviceguard has built in HA networking and in the latest versions can also monitor and respond to issues with RAC's cluster interconnect.
v) And of course HA doesn't begin and end with the database - if you want to cluster application components Serviceguard is a tried and trusted solution - I'm yet to come across anyone who has used Oracle's Clusterware for anything other than RAC.
HTH
Duncan
I am an HPE Employee