2540 array performance

I have a 2540 array and I'm monitoring it using CAM. I suspect the write speed to my volumes are slow. I've looked at the performance statistics in CAM, but it doesn't seem to give me the write speed. Below is a sample of the statistics I'm getting
Timestamp: Fri Mar 06 12:33:33 GMT 2009
Total IOPS: 416.83
Average IOPS: 157.08
Read %: 90.74
Write %: 9.25
Total Data Transferred: 14359.20 KBps
Read: 14065.17 KBps
Average Read: 4376.78 KBps
Peak Read: 16570.53 KBps
Written: 294.03 KBps
Average Written: 126.59 KBps
Peak Written: 294.03 KBps
Average Read Size: 271.64 KB
Average Write Size: 26.05 KB
Cache Hit %: 24.08
Does anyone know how I can determine of my write speed is slow or not.
Thanks

To determine if the array is running slowly, you need to really check on the servers.
Run 'iostat -zxn 60' (assuming Solaris)
The thing you're looking for is a high busy %age, high service time and a large number of requests in the queue. Either on their own is not really sufficient to point at a performance issue but both together would show that all is not well.
There's usally a lot more to it than that but that can give you a fair indication of potential problems.
You're showing quite high transfer sizes (writing avg 26K and reading an avg of 272K) - what profile did you use to create the volumes you're accessing here?

Similar Messages

2530/2540 array cache size vs. 6140

Greetings,
Needing some new storage we are considering the new 2540 array. Everything looks great except the cache size is up to 1GB. 512MB per controller. We are currently using 3510 storage arrays with dual controllers and 1GB cache per controller.
So at this point I'm thinking we should opt to go with a 6140 instead.
We've ruled out the 3510 just because we'd like this storage to take us 3-4 years down the road.
Any thoughts on this?
Thanks.

IMHO the 2540 is placed as "low cost entry" storage system, midrange is 6140/2GB and 4GB. everything higher: 6540/9990 ;)
i personally prefer ST6140/2GB as "entry" (better scalabality, nice IOPS)
-- randy

Array performance monitoring

I need some software to monitor Sun's arrays (2540/6140). The software should
save all the historical data so in case of any performance problems we can see
its performance day/week/month ago.
Does Sun has any such software ?

CAM is free of charge.
We sell also Santricity (I said sell, yes, this isn't free of charge, there is annual contract to use Santricity) BUT only under specific criteria which can certainly be explained by your Sun Sales representative.
You will certainly hear people saying that Santricity is better than CAM, or CAM is better than Santricity. Actually, there are different, but we work hard to implement the same functionalities in CAM than what we have in Santricity. We have a number of bug and RFE (request for enhancement) and CAM will even become better in the future, all this can be done because we own CAM and we develop CAM, which isn't the case for Santricity.
CAM comes by default with all new ST6000 and ST2500 arrays and again, at free of charge.
If you need any other information about one or the other, do not hesitate to contact your Sun Sales representative.
Regards

Copying arrays, performance questions

Hello there
The JDK offers several ways to copy arrays so I ran some experiments to try and find out which would be the fastest.
I was measuring the time it takes to copy large arrays of integers. I wrote a program that allocates arrays of various sizes, and copy them several times using different methods. Then I measured the time each method took using the NetBeans profiler and calculated the frequencies.
Here are the results I obtained (click for full size): http://i.share.pho.to/dc40172a_l.png
(what I call in-place copy is just iterating through the array with a for loop and copying the values one by one)
I generated a graph from those values: http://i.share.pho.to/049e0f73_l.png
A zoom on the interesting part: http://i.share.pho.to/a9e9a6a4_l.png
According to these results, clone() becomes faster at some point (not sure why). I've re-ran these experiments a few times and it seems to always happen somewhere between 725 and 750.
Now here are my questions:
- Is what I did a valid and reliable way to test performances, or are my results completely irrelevant? And if it's not, what would be a smarter way to do this?
- Will clone be faster than arraycopy past 750 items on any PC or will these results be influences by other factors?
- Is there a way to write a method that would copy the array with optimal performances using clone and arraycopy, such that the cost of using it would be insignificant compared to systematically using one method over the other?
- Any idea why clone() can become faster for bigger arrays? I know arraycopy is a native method, I didn't try to look into what it does exactly but I can't imagine it's doing anything more complicating than copying elements from one location in the memory to another... How can another method be faster than that?
(just reminding I'm copying primitives, not objects)
Thanks!
Message was edited by: xStardust! Added links, mr forum decided to take away my images

yeh, everyone thinks that at some point. it relies,
however, on you being perfect and knowing everything
in advance, which you aren't, and don't (no offence,
the same applies to all of us!). time and time again,
people do this up-front and discover that what they
thought would be a bottleneck, isn't. plus,
the JVM is much smarter at optimizing code than you
think: trust it. the best way to get good performance
out of your code is to write simple, straightforward
good OO code. JVMs are at a point now where they can
optimize java to outperform equivalent C/C++ code
(no, really) but since they're written by human
beings, who have real deadlines and targets, the
optimizations that make it into a release are the
most common ones. just write your application, and
then see how it performs. trust me on this
have a read of
[url=http://java.sun.com/developer/technicalArticles/I
nterviews/goetz_qa.html]this for more info anda chance to see where I plagiarized that post from :-)
Thanks for that link you gave me :)
Was usefull to read.
About time and money of programming, that is not really an issue for me atm since i'm doing this project for a company, but through school (it's like working but not for money).
Of course it should not last entirely long but I got time to figure out alot of things.
For my next project I will try to focus some more on building first, optimizing performance later (if it can be done with a good margin, since it seems the biggest bottlenecks are not the code but things outside the code).
@promethuuzz
The idea was to put collection objects (an object that handles the orm objects initialized) in the request and pass them along to the jsp (this is all done through a customized mvc model).
So I wanted to see if this method was performance heavy so I won't end up writing the entire app and finding out halve of it is very performance heavy :)

LabVIEW array performance: Two loops operating on the same massive array

Yes there has been a lot said and left unsaid about LabVIEW performance in the past - but I have to ask the following:
Is it possible to have two seperate while loops act on the same data without making copies?
Say I have a 1GB array of doubles which I have wired to a shift register on one loop. With a little bit of effort (by using in-place and chasing the dots a bit) it would be possible to operate on that data inside that loop without making copies.
But what if I need to also access that same array in a different loop (which is running at a different rate). I basically need two loops' shift registers to point to the same data. At the moment the only way I can manage this is to use DLL calls which destroys dataflow but al least makes no copies. It also means I cannot use LabVIEW code but have to resort to C coding.
Data corruption is no issue as the two loops, at any instant, operate in different locations of the array.
Any smart ideas that will work?

drclaw wrote:
Hi AnthonV,
A Functional Global approach might suit your needs. This will allow you to keep you data in a shift register, but access/modify parts of the data from other VIs or loops.
Another option is to use a single element queue to store the array to achieve a similar result.
Both these options and more techniques are located at
http://zone.ni.com/devzone/cda/tut/p/id/3625#toc0
Hope that helps
Good suggestion. Functional Globals also go by the name "Action Engine" as I discuss in this Nugget on Action Engines.
Put all of your number crunching code inside appropriately named Actions of the AE and that way all of the work CAN be done "in-place".
Have fun!
Ben
Ben Rayner
I am currently active on.. MainStream Preppers
Rayner's Ridge is under construction

Attaching 2540 array to Solaris 8

Hello,
We are in a pickel. Currently, we have a 3511 connected to a V880 running Solaris 8.
(SunOS agatlsun01 5.8 Generic_117350-43 sun4u sparc SUNW,Sun-Fire-880).
And this array is not worth fixing any more.
We are purchasing a 2540 as a replacement.
We run Oracle 9i and 10g databases on this server.
I've been told that the 2540 is not supported by Solaris 8.
We would like to have both arrays connected to the same box to allow us to migrate the 5 databases off the 3511 and on to the 2540.
Is anyone aware of patches or work-arounds that would allow us to attache 2540 to the V880 running Solaris 8?
If this is not possible we will have to attach the 2540 to another server, migrate the databases to this temporary server, then attach the 2540 back to the orignal server. Seems like a lot of unnessary work.
Thanks.

You can find the supported components here:
http://sunsolve.sun.com/handbook_private/validateUser.do?target=Systems/2540/components
The only other problems, I can think of would be with:
1. The QLC driver
2. The CAM software (verified this is supported under Solaris 8 SPARC 4/01)
So that leaves the drivers.

Arrays performance - 1d vs 2d

Which solution is faster:
int[][] tab = new int[100][100];
tab[x][y] = c;or:
int[] tab = new int[10000];
tab[y*100+x]=c;I know that the second solution is recomended, but it requires one multiplication.
In the first solution compiler can store pointers to second dimension table in first dimension. Pointers have size of 2^n. Integers in second dimension are also of size 2^n. So theoretically compiler may produce byte code that will need no multiplication but only very fast bit-shift operations.
My question is - how does the compiler cope with 2d arrays.
I'm posting my question here because I don't understand bytecode so well. I'm looking for theoretical answer, not benchmark result for specific mobile device.

If I understand right, the problem with that code for the distance matrix is that I dont want to recalculate every row with each iteration, just merge 2 rows from the original distance matrix into 1...and also I cant pick and choose which rows I am merging. There is another function that tells me which rows I have to merge. I could do it with a regular 2d array, but then I need to either shift all the rows each time i merge/remove a row, or I have to keep track of which rows in the matrix i am still using...that would be more complicated than just doing an arrayList of arrayLists I think.
Thanks for the pointer on the arrayList of arrayLists....i'm gonna try to use that for the distanceMatrix. Do you think I lose anything (efficiency or memory) by implementing the 2d arrays the way I have, rather than the arraylist for the expressionMatrix?

Vector vs Array performance?

Hi, I've heard some ppl says that Vector is slower than Array. Does anyone know what's the different in speed between using an array and a vector?
Say with the same algorithm, the one uses array take 10s for 1mil iteration, how long with it take for vector?
Thanks...

For what it's worth adding 10K strings to various collections and accumulating 1000 results, this is what I saw:
Vector() 0.69/0.53:
ArrayList() 0.84/0.69:
HashSet() 2.15/0.36:
Vector(SZ) 0.70/0.53:
ArrayList(SZ) 0.25/0.43:
HashSet(SZ) 1.56/0.52:
LinkedList() 0.86/0.60:
A slight win to Vector unless the collections have their initial size set, in which case ArrayList comes out ahead.
Things are a little bit different using 500K strings and doing fewer runs (in both cases I didn't time the first two runs):
Vector() 46.20/7.77: 50 38 49 38 50 63 49 38 49 38
ArrayList() 51.60/6.00: 50 48 51 47 51 69 52 49 51 48
HashSet() 662.30/31.28: 648 652 652 651 652 756 652 653 652 655
Vector(SZ) 14.90/0.30: 15 14 15 15 15 15 15 15 15 15
ArrayList(SZ) 33.20/7.21: 26 41 26 40 26 41 26 40 26 40
HashSet(SZ) 520.30/1.73: 523 519 520 518 519 521 519 523 522 519
LinkedList() 345.90/31.44: 336 339 332 336 440 336 334 337 332 337
If the initial capacity is set Vector was a clear winner for collectons of this size.
Edited by: pbrockway2 on Jan 10, 2008 11:16 AM
I have no fscking idea why Vector should appear to get slower when the initial size is set to 10K. But, there you are.

StorageTek 2540 and 2501 performance

I'm doing some work with a dual controller 2540 with 2501 expansion trays attached and would like to identify what the maximum throughput of the arrays should be.
This is how I understand the architecture:
Each 4Gbit FC host port can be active concurrently providing up to 400MB/sec per port, total of 1600MB/sec between the hosts and the array.
Each controller has a PCI-X bus between the host board interface and the SAS I/O controller, capable of 1GB/sec per controller.
The "SAS I/O controller" in each controller connects to it's local "SAS Expander" using a 2x wide SAS channel = 6Gbit/sec, and also connects to the other controller's SAS expander using (another?) 2x wide SAS channel.
The SAS expander in each controller has a separate single lane, 3Gbit connection to each of the 12 internal disks.
Each disk has a connection to the two SAS expanders, but only one is active (?). Total connection per disk to the rest of the array is 3Gbit?
There is a 4x wide connector from each SAS expander to the SAS drive port (to connect to a 2501). This has a throughput of 12Gbit/sec?
If the above is correct (and please feel free to put me right if I've got something wrong!), what happens in the 2501? From reading the documentation on the 2501, it would appear that the connection between the 2501 and the 2540 is 3Gbit (per controller). Is this a significant bottleneck if two 2501 expansion trays are added?
I got most of the above from http://shop.tisource.ch/pdf.php?id=1619793 and trying to match it with jmiller's comment in https://opensolaris.org/jive/thread.jspa?messageID=258953
Any insights appreciated.
Thanks
JR

The same result here for a 2540 attached to a Sun M4000 (with qlogic-hbas) running latest Solaris 10 8/07 with latest Patches. There is an 3510 attached also. The mpxio-device for the luns of the 3510 are working well.
The command "mpathadm-command show mpath-support libmpscsi_vhci.so" shows that the 2540-array (LCSM100_F) is supported. Nevertheless the mpxio-devices for the 2540 are not shown up.
I will making a sun-call now.
Christian

StorageTek 2540 : set duplex controller using CLI (sscs)

Hello all,
StorageTek 25xx disk enclosure software documentation mostly relies on the CAM graphical interface for performing tasks.
However, for those who wish to use the command-line interface (CLI), several tasks are not - or not visibly - documented.
We have a few StorageTek 2540 arrays which initially shipped with only one controller board per array. Then a second controller was added to each array.
When the second controller is installed, it has all 3 LEDs lit (green, orange and blue) and is seen as "failed" with the "sscs" CLI command.
Using the GUI, the "set array to duplex mode" menu, followed by a power cycle of the array, effectively enables the second controller.
So the questions are:
- How does one enable duplex controller mode on a StorageTek 25xx using the "sscs" CLI command ?
- More generally, is there any advanced CLI documentation available, the CLI "quick reference" and sscs man page being insufficient for this kind of task ?
Thanks in advance for your answers.

hl_fr wrote:
Nwi, thanks a lot for the information you provided.Your welcome.
This was the only missing link for complete array commissioning using only the command line.
IMO, this option should be in sscs. Actually, this is a service CLI, this isn't used for array management. And service actions are available in Service Advisor, that's why this is in the 'service' CLI.
Also, the "failed" state of the secondary controller before the array is set to duplex is confusing.
I understand, do no hesitate to open a call with us and request a RFE, this will be evaluated.
So to summarize for those interested, the main steps for setting up a Storagetek 2540 array from scratch are:
- Do the cabling stuff.
- Install the CAM software on a chosen host.
- Using the serial cable an a virtual terminal, set up the IP parameters on the primary controller.
- Using sscs(1M), declare the array in CAM with the primary controller IP address.
- Set the array to duplex mode using the undocumented command you provided:
/opt/SUNWsefms/bin/service -d <array name> -c set -q redundancy -t duplex
Note: as per the GUI online help, this requires a power cycle of the controller array.
- Using sscs, set up the secondary controller IP parameters.Don't forget that Service Advisor (GUI) does a number of other actions, you also need to set the battery age for the new controller, you need to generate an new inventory, check the f/w, etc...
Therefore, Service Advisor is what should be used for such service actions.
Regards

Installing Sun firmware on Seagate disks for StorageTek 2540

Hello,
I recently purchased a StorageTek 2540 array that has several open drive slots. As luck would have it, I also have several Seagate ST3146356SS SAS drives sitting on the shelf that according to the System Handbook are one of the models of drives supported in the STK2540 (Sun part #390-0422).
So I popped them in the chassis, and they show up immediately and are available for use, which is good.
The problem is, these drives have standard Seagate firmware on them (version 0006) rather than Sun firmware. As such, they show up with a different capacity than the Sun 146GB drives that are already in the chassis (which happen to be Hitachi, rather than Seagate drives).
I would like to update the firmware on these drives to the official Sun version of the firmware, but I cannot see how to do that. I tried to use the Install Array Firmware Baseline wizard, but when it analyzes the drives, it tells me for the Seagate drive:
Current Firmware: Tray.85.Drive.06: 0006
Baseline: Tray.85.Drive.06: Not Applicable
In the past I owned a bunch of Sun T3 arrays, and I used to replace standard manufacturer firmware with Sun firmware all the time, and it was very useful. I'd like to be able to do the same thing here. I also have a SAS controller in my PC, and could use Seagate's Seatools or DriveDetect tools (Windoze) or Seagate Enterprise CLI (Linux) to update the firmware, I believe.
It looks like I have the Sun firmware on my Solaris server (the /opt/SUNWstkcam/share/fw/images/disk/D_ST314656SSUN146G_0A1C.dlp file included in CAM 6.7).
Is it safe to grab that file and use it to update the drive firmware? Or is there a better way?
I realize that burning firmware to a non-Sun drive is not supported, and may render the drive(s) useless, but as I am not using the drives for anything else today, anyway, I'm willing to take that risk.
Any suggestions would be appreciated.
Thanks,
Bill

FYI, after some digging around, I discovered the "sscs modify firmware" command and tried that, but it did not work:
61# sscs modify -a ss2540 -c Tray.85.Drive.06 -f -o -w \
-p /opt/SUNWstkcam/share/fw/images/disk/D_ST314656SSUN146G_0A1C.dlp \
firmware
Analyzing Firmware
Incompatible firmware image. Skipping Tray.85 Slot.6
Firmware install failed.I'm guessing that since the non-Sun drives show up as model # ST3146356SS instead of ST314656SSUN146G, the firmware is deemed incompatible by the CAM CLI.
So I'm still looking...
Thanks,
Bill

EVA performance with ESXi 5.1

We are having problems with some of our Linux VMs going read only and Windows VMs slowing right down on our P6000 EVA. The problem tends to appear during a backup window, but sometimes during the working day under heavy periods of activity. It seems to be storage performance related. ESXi 5.1 is using round robin and there are no network issues.
Has anyone tried these advanced ESXi settings which are reported to improve performance?
(From: http://arritdor.e-wilkin.com/2012/05/vmware-esxi-and-hp-3par-storage.html)
DiskMaxIOSize = 128
QFullSampleSize = 32
QFullThreshold = 4

I think it might be beneficial to take a look at the array performance statistics. The evaperf can be found at c:\Program Files\Hewlett-Packard\EVA Performance Monitor if it is selected during CommandView installation. Some useful stats to collect are:
evaperf cs –cont 60 -dur 3600 -csv -ts2 -sz xxxx-xxxx-xxxx-xxxx > controller.csv
evaperf hps –cont 60 -dur 3600 -csv -ts2 -sz xxxx-xxxx-xxxx-xxxx > hostport.csv
evaperf vd –cont 60 -dur 3600 -csv -ts2 -sz xxxx-xxxx-xxxx-xxxx > vdisk.csv
evaperf pdg –cont 60 -dur 3600 -csv -ts2 -sz xxxx-xxxx-xxxx-xxxx > diskgroup.csv
evaperf pda –cont 60 -dur 3600 -csv -ts2 -sz xxxx-xxxx-xxxx-xxxx > disk.csv
where xxxx-xxxx-xxxx-xxxx is the world wide name of the array e.g. 5000–1FE1–5000–A9F0. You can probably find it on the controller LCD or some serial number label sticker at the back.
Because window server command prompt does not support running command at the background (at least I do not know how) unlike hp-ux server, you will have to open several command prompt window and run these command at the same time. Then probably zip and posted it somewhere so that we can all take a look.
the 60 here means we make one collection every 60 seconds, and -dur 3600 means for a duration of 3600 seconds which is one hour. For a start, we do not want to collect huge amount of data. We probably want to collect one set when array has no problem, and one set when array has problem, then make a comparison.

Disk Array volumes showing up on servers

I have a StorageTek 2540 array with 2 controllers (A & B). I also have 2 servers T5220 (connected to controller A) and a T2000 (connected to controller B)
I have created several volumes on the array some pointing to the A controller, some to the B controller.
When I run the format command on the 5220 (A controller) all of the volumes show c2t0d0...d3.
From the T2000 (B controller) the format command shows c2t2d0..3.
If you try from the 5220 (A controller) to format a volume that is pointing through the B controller, it won't work. That's great no destroying an existing volume.
I know which volume points to what server, that isn't a problem yet.
My question is:
Is there a way on the Array to only show the volumes pointing to each controller individually instead of together?
thanks
Bill

Hi:
Try each of resolution steps 2, 3, 4 and 5 from the link below.
Normally #3 or #5 will fix the problem you are having...
http://support.microsoft.com/kb/314060

Xserve-RAID array intialising

An established array on my RAID has sudddenly begun the intilising process. Is this normal? Why would it be doing this now - it was initialised when I set it up so why now?
Also - why does the initialising proces take so incredibly long?!

Initializing? that sounds like a problem. Unless you mean either scrubbing/conditioning or rebuilding.
The former can be initiated via RAID Admin, and makes the XServe RAID check every block on the disk.
The latter may happen if a disk fails and the data needs to be replicated/rebuilt on a spare drive.
Initializing, as its name implies, involves setting up an new array. This doesn't usually (ever?) happen on its own, but only at the behest of someone using RAID Admin to manipulate the array.
As for the duration - if you're using RAID 5, creating the array performs a scrub of each disk to verify its integrity, as well as creating the initial parity data. If you enable the background option, the array is available for use immediately, even while the scrub is progressing.

2540 - more then 32 mappings to the host ?

Hello,
we decided to chunk the whole 2540 array space into 100 GB chunks. We got about 40 LUNs.
But we cannot map all to the host - only first 32 LUNs.
Does anybody know why ?

HI,
It is the limitation of the storage and you cant map more than 32 luns from the storage.
you wont get the option of lun number greater than 31 and 31 will be a phantom.

2540 array performance

Similar Messages

Maybe you are looking for