Difference Between TimeFinder and TimeFinder Clone?

Question

a) What is Timefinder?
b) What is Timefinder Clone?
c) Example of using Timefinder?
d) Example of using Timefinder Clone?

I'm a newbie. And although I've been an HP-UX admin for many years this is the first time in the role of an EMC Storage Admin.

And I do have books to read. But I'd like to here from the source.

xe2sdc · Accepted Answer

Good old TimeFinder (TF/Mirror) and TimeFinder Clone (TF/Clone) gives you the same result (snapshot of your data) using two totally different (and for some aspects, opposite) mechanisms.

With both products you have 2 different phases .. a moment when you create consistency (point in time copy) and a phase while the box copies data from source to target volumes.

With TF/Mirror you have first to copy data (establish) and then create PIT (split).
With TF/Clone you create sessions at first and later set PIT and start copy with a single command (activate).

They both tend not to overload your box however since TF/Mirror uses internal mirroring algorythms, it tends to be more aggressive with respect to host performances while TF/Clone tends to reduce the pressure on the backend. But note that your specific environment (number of clones for any given source, backend layout, cache size, box type, code version) may give VERY different results.

dynamox · Answer

Michael,i would recommend you post your question in either Symmetrix or Solutions Enabler section of the forum. There is a lot more traffic there. In the mean time have you had a chance to download 'EMC Solutions Enabler Symmetrix Timefinder Family CLI' guide ..it contains very good description and examples. Welcome to the forums.

xe2sdc · Answer

Making your every wish come true!

xe2sdc · Answer

Just to make a long story short ...Timefinder is (right now) the name for a whole line of products ...You have 'TimeFinder/Mirror' (AKA good old TF) but also TF/Clone and TF/Snap. They all share the same background idea: give you a copy of your beloved data. They have different behaviour and requirements, they have different limits and possibilities. However whenever you need a copy of your data inside your local storage, you ask TimeFinder.What gives you any TF member ?? They all will give you a different set of devices (scsi disks) that contain a copy of your production data, hopefully a consistent PIT image of your data. Note that since you are asking the storage to build the copy, you are offloading the workload from your servers to the storage. Hopefully your hosts will never notice what's going on (even if it may happen sometime).Difficult to give examples .. If you come here in Italy I can show you a customer using TF/Clone to validate 95 Oracle instances at DR site ... Or using TF/Mirror to backup a 10TB database without shutting down the database I'm pretty sure there are a lot of other bright examples of what TF gives you.A note on the name ... we have a lot of products .. SDDF, SRDF and so on .. However I find those names a little cryptic .. TimeFinder is indeed a great name since it's WYSIWYG .. Time Finder .. Find time you don't have .. You need to backup your database. But your need to take orders 24/7 .. How can you backup 10Tb of data while keeping the database open and being able to process orders from customers ?? Yes I know you have 'begin backup/end backup' but it will hurt your customers. And how long will it take to backup 10Tb of data ?? .. So you have a brand new answer .. TimeFinder. You keep your DB in hot backup mode just for the time needed to create a consistent PIT image INSIDE the storage. And when it's done, you can 'end backup' and have your DB run again at full power. And it will take minutes, not hours (or days).

Allen Ward · Answer

The other thing you need to pay attention to is how the two technologies (mirror and clone) handle the relationships internally.

Many admins who have been using Symms for some time seem to prefer mirror as the default. I haven't been using Symms for that long and find clone to be much more useful (in our specific environment).

Mirror has quite a potentially limiting number of copies you can maintain without some fancy shuffling, while clone is much more liberal in the number of copies. This can become even more critical if you are using SRDF for those same devices.

Also I would like to clarify a statement on Clone copying the data after the activate is issued. This is a user controllable function as we precopy the data prior to activation. This makes the clone work a bit closer to a mirror copy.

Finally, there was TF/Snap mentioned, but not described at all. I know you didn't ask specifically, but it won't take long to explain it in the mix. Mirror and Clone both create separate "physical" copies of your source device. SNAP creates a virtual (pointer based) copy of your data. This could potentially affect performance of the source device while it is being accessed, but there are advantages to being able to create multiple snaps of a 1TB database file system without consuming multiple TBs of capacity.

Everything has it's place, and your specific application of the technology will drive out which "flavour" is best for you.

michaelsteele1 · Answer

Thanks Stefano, especially for the third reply.

xe2sdc · Answer

Since code 5671 yep ... You need also S.E. 6.3 on the hosts I think .. however works great with 6.5.x

RRSiemers · Answer

Most my experience is with mirror...

I know that BCV mirrors maintain an association with the source disk to keep track of which sectors have changed, on both sides of the mirror. This allows for incremental re-syncing of the mirror to save time by only copying the deltas. Like wise, in the event of our source data going bad, an incremental restore can get you up and running faster than a full restore.

Do the clones also support incremental resyncs?

Message was edited by:
RRSiemers

Symmetrix

Difference Between TimeFinder and TimeFinder Clone?

Was this post helpful?