Advice on new hard drives and slowness issue

[Bigsease30]

Hi All,

 

It's been a while since I last visited the forum. I see all is still going strong which is awesome!

 

I am currently in the process of switching to a new server. I am using a PowerEdge R510 which has 12 bays, a PERC 700 card and is using Server 2012 R2 Essentials and have currently acquired a brand new PowerEdge T630 with 18 drive bays that will be using a PERC HBA330 and will have Server 2016 Essentials. I will receive it this Friday, hopefully. The R510 has each drive setup in a raid0 configuration. I have hated this setup from the start so that's what I ordered the HBA instead. I currently have 12 - 4TB Seagate ST4000DM000 drives with only about 1.5 TB left of free space. I would love to start moving to the 6-8TB drives but I am not sure what would be the best route. Something else that I have been experiencing is a slowdown in write speed. Sometimes when I start a file copy from my PC to the server, it will stall for up to a minute then start transferring. Other times it will start and avg 112mb of the LAN and then just stop for a min then start transferring again. I am not sure if that is the drives causing the issue or the PERC card since the drives are in RAID0. 

 

Any assistance or advice would greatly be appreciate.

[Spider99]

Hi

 

a few things to consider

 

on the slow copy - do you have windows search enabled to index the drives - have seen this cause a few issues - mine was more creating a directory - but turning it off generally improved performance - might be worth a try

 

having low space across 12 drives will limit options for placement or make it take longer to decide where to put the data

 

do your drives spin down when not in use - i have mine on an aggressive 10 min spin down and do get pauses of 10-15 seconds while a drive spins up etc - but your pauses sound significantly longer

 

HDD do get significantly slower when they are nearly full (writing to the centre of disk) 

 

those HDD from seagate are notorious for failing in high usage environments (i have lost 8 out of 12) - but as you are not commenting on them i guess you got a good batch

 

Are there any errors in the event log/ Perc firmware/ smart info etc etc

 

to move the data - i guess all of it from one pc to the other - a) copy across the network pull individual drives and connect via usb or install in new pc and copy off the data to a new hdd c) move the whole pool to the new pc add some new disks and slowly remove the old drives via the UI - it will take a few hours per disk

[Bigsease30]

Hi Spider99, Thanks for the reply.

 

1) Windows indexing is turned off.

 

2) Drives are not set to sleep so they should not be spinning down. It really is a weird issue. sometimes it will freeze several times during a single file transfer, granted most transfers are around 1.4gb. (mostly HD movies). I only started noticing this issue when I moved to the R510 and 4Tb drives.

 

3) I have seen the slow downs even when the pool was 1/2 full.

 

4) Yes, When I was using all Seagate 2TB drives, I started to lose one about every 3 months. I have read that it was a bug in the FW that was causing this to occur. Most drives were still under warranty so I still have some new ones here in my storage cabinet. I have been using these 4TB drives for over a year and a half now and have not had a single drive fail....yet.... I couldn't pass us the deal I got for all 12.

 

5) I don't see any errors listed in the PERC FW nor OpenManage. I have not been able to use the SMART feature due to them being on a raid0. The controller doesn't allow true pass-through of the data so I gave up on that a while ago. Hence why I ordered an HBA this time around.

 

6) I wish moving the data would be as easy as pulling a disk and coping the information. I have never been successful in pulling a single disk and reading the contents of the raid0 disk. I ordered 2 - 10GB network cards from ebay with a cable for only $37.95. Hopefully this will cut the time down to a few days instead of a few weeks. I will also be creating a new pool on the new server with all formatted disks. (Hopefully new ones.) I am leaning towards the WD Red Pro 8TB disks.

[Spider99]

http://wiki.covecube.com/StableBit_DrivePool_Q4142489

 

thats the quick and dirty method which can be easily modified for your situation

 

As you do not have smart data - there could be an issue with a disk or two - i would scan them when you put them in your new box with the hba

 

with the older perc's you can flash them to IT mode (i.e a hba) not sure about your 700

[Bigsease30]

Thanks,

 

I will give the quick and dirty method a try.

 

Yeah I looked into flashing this perc card before I configured that server and found that it is not possible. Oh well, out with the old and in with the new. I am expecting 8 x 6TB WD Red Pro's to be delivered tomorrow and the server on Friday. Wish me luck!

[Christopher (Drashna)]

One thing that may be helpful.... if this is over the network ... there are some tweaks that you can make to help performance. 

 

In device manager (on both systems), make sure that any setting with "offload" or "checksum" in the name is turned off. ESPECIALLY if this is an onboard NIC. 

Try toggling "Green ethernet" if it's an option, and try disabling jumbo frames. 

 

Also, on both systems, try running the following from elevated command prompts. 

"netsh int tcp set global autotuninglevel=highlyrestricted"

[Bigsease30]

Awesome Christopher, Thanks for the advice.

[Christopher (Drashna)]

You're very welcome.