1819836 Members
2799 Online
109607 Solutions
New Discussion юеВ

Re: DFO or Diskeeper

 
SOLVED
Go to solution
Tim Nelson
Honored Contributor

DFO or Diskeeper

Looking for anyone's experiences with HP's DFO or Network Dynamics' Diskeeper products.

I was also testing the HP show_alpha.exe program that reports on disk frag. Noticed immediately that it does not ODS-5 filesystems.

Any comments would be greatly appreciated.

Thanks !
19 REPLIES 19
Bill Hall
Honored Contributor
Solution

Re: DFO or Diskeeper

Tim,

I have never used DFO so I can't comment on it. Seven or eight years ago I replaced Diskeeper here with a combination of Raxco's PerfectDisk and UltraDisk on VAX and Alpha.

I was directed to do the same at my prior employer. The decision there was made after a corporate evaluation of Diskeeper, PD/UD and I believe the very first release of DFO.

My personal opinion of Diskeeper vs PD/UD has always been:

Diskeeper looks "pretty" but doesn't do much for you. PD/UD doesn't look as "pretty", but it really does something for you.

PD doesn't just defrag files, it can also attempt to move files to the most efficient location on disk for the type of file. UD helps minimize fragmentation and assists PD in moving and defragging open files.
Bill Hall
Karl Rohwedder
Honored Contributor

Re: DFO or Diskeeper

I use DFO on several sites for some years now and am quite happy with. Once setup, it runs on its own.

regards Kalle
Uwe Zessin
Honored Contributor

Re: DFO or Diskeeper

Didn't like DK either, but had good experience with Raxco's and DEC's products.
.
Tim Nelson
Honored Contributor

Re: DFO or Diskeeper

Any performance or overhead issues worth mentioning ?

There seems to be a big difference in price.
HP DFO = $800
Raxco = $8000

Thanks all !
Uwe Zessin
Honored Contributor

Re: DFO or Diskeeper

Unfortunately, I am not familiar with current product states, but Raxco's has a much longer history than DEC's. They (DEC) for a long time disclaimed the need for defragmentation until they provided defrag hooks in the XQP.
.
Antoniov.
Honored Contributor

Re: DFO or Diskeeper

My experience was in past.
Raxco Diskeeper worked very very good! High performance, high stability.
Today I can't compare DFO with Diskeeper.

Antonio Vigliotti
Antonio Maria Vigliotti
Uwe Zessin
Honored Contributor

Re: DFO or Diskeeper

Raxco's product was called Perfect Disk, not DK.
I agree that PD was very good.
.
Bill Hall
Honored Contributor

Re: DFO or Diskeeper

Tim,

I don't know for sure how competively priced Raxco is in the market place. I would expect them to be more competive than that. You may be looking at apples and oranges.

My guess is that the price you have from Raxco is for their "Performance Suite". The Performance Suite includes PerfectDisk, UltraDisk, PerfectCache, PerfectTune and I think it even includes PerfectFile. You should be able to get ala carte pricing from Raxco depending on what your real needs are.

If you are running a VAX environment, PerfectCache will blow-away the VIOC. XFC is more competitive performance wise with the PerfectCache. PerfectCache provides more management knobs to tweak than XFC. PerfectCache (and UltraDisk) must be licensed and run on every node in a cluster. If you are interested in dynamic system tuning, then it would make sense to license every node in a cluster for PerfectTune. PerfectDisk and PerfectFile can be licensed and run on a single cluster member.
Bill Hall
John Gillings
Honored Contributor

Re: DFO or Diskeeper

Tim,

Regarding Uwe's remark:

"They (DEC) for a long time disclaimed the need for defragmentation until they provided defrag hooks in the XQP. "

Well, it was more like every time a new version of VMS (as it was then) was released, lots of disks were broken because the defrag makers had made invalid assumptions about how the XQP worked. The defrag hooks were implemented to protect customers. A corporate defragger was also created because (in a rare flash of marketing sanity!) someone realised there was a market for it and DEC could make some money.

Some of us still disclaim the need for defragging disks. It should be possible with a small investement in planning and management to avoid the need to defrag files, simply by preventing them from fragmenting in the first place. Sometimes it's as simple as setting reasonable extend sizes.

It should also be realised that the OpenVMS file system needs to be VERY fragmented before there are any serious performance issues, so a little fragmentation isn't a problem. Indeed, for some workloads, a fragmented disk is GOOD (for example, many small files being created and deleted).

Further realise that OpenVMS is a multiuser operating system, so there are many processes using the disk at the same time. Thus, the one place you can be almost certain a disk head will NOT be is where you left it! A lot of the myths about keeping disks defragged is from single user systems and less sophisticated file systems.

Before spending big bucks on a license for a defragger, please find out what (if anything!) is the REAL problem. Not only can you save on the cost of the software, you can also save the resources need to run the defragger.
A crucible of informative mistakes
Uwe Zessin
Honored Contributor

Re: DFO or Diskeeper

It was certainly a noble gesture to protect the customers ;-)

However, when you buy a piece of software from a vendor you usually neither have access to the source nor the time to 'tune' these programs. And, trust me, I have see quite a number of bad packages for VMS. I was able to get a little look at the source code of some programs and, gasp, it is amazing how bad a program can be and still work, somehow.

Another good thing was that defragmenting the disks did cut down the time required for backups, so this saved some time spend on operators. I am talking about the time when TU77 and TU78 tape drives were used!
.
Ian Miller.
Honored Contributor

Re: DFO or Diskeeper

DFO works fine for me but I agree that you should look to see if fragmentation is actually causing you a problem. A disk can be fragmented and have no impact. You need to look at the performance of your system running your application not how fragmented or otherwise your disks are.

____________________
Purely Personal Opinion
Jan van den Ende
Honored Contributor

Re: DFO or Diskeeper


Many, many moons ago it was explained to me how investing in 'some' memory can make file fragmentation into a near non-issue.
Set ACP_WINDOW to 255 (max) and you have Cathedral Windows, meaning even a very fragmented fileheader is read into memory in its entirety, preventing Window Turns.
Still each fragment will have to be read, but locating it is no longer an issue.

Since then, we never felt the need for defragging.
Maybe just because we are not demanding enough? :-)

But, as so often, it is far easier to explain to management the need for defragging as they know it form their home toy, as to explain why for serious systems the balans of cost vs gain tilts the other way...

FWIW

Proost.

Have one on me.

jpe
Don't rust yours pelled jacker to fine doll missed aches.
Robert Atkinson
Respected Contributor

Re: DFO or Diskeeper

Tim, we've got lots of experience with Perfectdisk - or Rabbit7, if you're memory goes back that far, and with Diskeeper.

Diskeeper is supported in the UK by Executive Software, and all I can say is DON'T GO NEAR IT!

Diskeeper comes with a module called FRAG_GUARD that stops fragmentation at source. This broke on our HSG's (killing the system first of course) and after 2 years of complaining, was never fixed.

In the meantime, we had endless battles with ExecSoft over why DK wasn't defragging certain files. They kept blaming it on the files being open, even though I could prove the disks we're mounted privately to the DK process.

In the end, we went back to Raxco. Once set up, you just forget about PerfectDisk and enoy life.

I did try DFO, but it wasn't as rich as PD and didn't seem to make as good a job of defragging the disks, but that was a few years ago now.

BTW, the reason we went from PD to DK is because after a VMS upgrade, we found PD was moving the Boot Block around even though it was marked NOMOVE, and lost confidence in it. This is the only problem I've ever known.

Tim Nelson
Honored Contributor

Re: DFO or Diskeeper

Thanks again for the info/experiences.
So far here are some other items that we found.

Raxco's PD does not support multi-path on 7.3-1 (not sure about 7.3-2 or higher)

DFO does not do open files.
Raxco does do open files.

I will continue to update as we progress with our testing.

David B Sneddon
Honored Contributor

Re: DFO or Diskeeper

Tim,


Raxco's PD does not support multi-path on 7.3-1 (not sure about 7.3-2 or higher)

I would be concerned about this. See the following
http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=886370


DFO does not do open files.

This can be annoying but seems to me "safer"


Raxco does do open files.

I would be concerned about this also.

(Purely personal opinion and having had bad
experiences with earlier third-party defragmenters.)
Dave
Jan van den Ende
Honored Contributor

Re: DFO or Diskeeper

Tim, & David,

No support for multipath looks like a potential issue to me as well.

But on the topic of Raxco's open file handling:

I remember the introduction of this at a Decus, (ISTR it was the 1999 SanDiego, but not sure there) in a double (2-hour) session.
A seizable part of the audience were Digital OpenVMS engeneers, who were VERY critical.
The second hour existed solely of questions by Engeneering, and answers by the Raxco techies. After this collective effort Engeneering concluded that they could not find fault, (although they would keep searching of course), and THEY initiated the big round of applause.

-- I have no info about ODS-5 issues.

hth.

Proost.

Have one on me.

jpe
Don't rust yours pelled jacker to fine doll missed aches.
Jim Hibbits
New Member

Re: DFO or Diskeeper

Regarding PD and Multipath:

PD does _not_ have an issue with Multipath.
UltraDisk _does_ have an issue with Multipath. UltraDisk is the component that allows PerfectDisk to defragment open files.

The issue with Multipath is coming to an end Real Soon Now, having required some cooperative effort from HP. This issue affected both the UltraDisk product and the PerfectCache product. I've spoken to the responsible engineer as recently as today, and it is believed that the resolution will be ready for public consumption this summer.

Disclaimer: I work for Raxco. And that was probably me at the 1999 event...
comarow
Trusted Contributor

Re: DFO or Diskeeper


I've been involved in the past as a customer of Diskeeper and more currently with DFO as an HP employee. I was recently involved with an elevation regarding DFO on initial runs of huge disks with a small cluster size. It provided a lot of insight into DFO engineering. A slight change in code resulted in huge performance improvements in this unusual situation, which will of course go downstream in all future DFO versions.

The DFO engineering was very interested in the results and we learned a great deal about file system performance. On a huge disk with small files a little bit of fragmentation is actually desirable as the disks seek free space.

But perhaps the most important aspect is that the engineering groups involved in all the storage products work together and thus are able to work cooperatively. And, they know what is coming down the "pike".

The ability to elevate to engineering any problems you experience is invaluable.

comarow
Trusted Contributor

Re: DFO or Diskeeper

I thought I should add, with XFC, most recent files will be in momory.
XFC is an aggressive read ahead program for caching RMS reads, with hooks for future writes.

Unlike other defragmenters it is part of VMS and will not expand npagedyn leading to other problems.

You're heavy hit files will be in memory.

Yet it is smart enough not to cache chached, products, like SGAs, or RDB caches.

The I/O rates have been exceptional.

And of course, while HP will work with third party vendors, you avoid all finger pointing for faster resolution to any problem.
Bob

Bob