Tag Info

Hot answers tagged

11

TRIM works at the filesystem level so as you're giving your KVM domains a raw block device then you need to enable TRIM from within the domain; your host can't know the domains filesystem utilisation without examining it. To enable this you need to ensure there's a discard='unmap' attribute added to the disk definition in the XML for the domain, this is ...


7

In addition to the other valid remarks: That particular drive, the Samsung 845DC is in the words of the manufacturer "designed for read intensive, <10% write content" and a write lifetime of 600TB which, depending on the IO profile of your VM's, may result in an early death, not covered by the 5 year warranty. Server SSD's are typically specified for ...


6

This all depends. If HP or IBM, I'd say use their respective drives. (just because) If Dell, probably use their drives... If you can't afford the Dell-spec'd disks, look harder. But refurbished Dell disks if you have to in order to save money and retain support. But also know that Dell PERC RAID controllers are manufactured by LSI, and LSI controllers ...


6

You can't use non-HP SSDs in HP ProLiant servers like this. Just because this worked on your G7 server doesn't mean it is okay for your Gen8 ProLiant servers. (basically, why buy enterprise gear, then cripple it with incompatible components?) Please see: 3rd party SSD drives in HP Proliant server - monitoring drive health or Third-party SSD solutions ...


5

The cached reads aren't from the underlying device. They aren't representative at all of the performance of the different devices, neither of which could come anywhere near that kind of throughput. Ignore that number.


4

I can speak on the specifics of what you're trying to accomplish. Honestly, I would not consider an entry-level HP P2000/MSA2000 for your purpose. These devices have many limitations and from a SAN feature-set perspective, are nothing more than a box of disks. No tiering, no intelligent caching, a maximum of 16 disks in a Virtual Disk group, low IOPS ...


4

The rule of thumb I use for disk IO is: 75 IOPs per spindle for SATA. 150 IOPs per spindle for FC/SAS 1500 IOPs per spindle for SSD. As well as IOPs per array also consider IOPs per terabyte. It's not uncommon to end up with a very bad IOP per TB ratio if doing SATA + RAID6. This might not sound too much, but you will often end up with someone spotting ...


4

The manufacturers have spent time validating OEM drives and possibly creating custom firmware to deal with compatibility/optimisation issues with specifically their RAID controllers. There is some value but it is very untangible. Some products simply won't accept non-proprietary drives. Also until very recently server grade SSD products simply were not ...


4

Your SSDs are likely healthy, but the HP Smart Array P410 RAID controller is not compatible with every SSD. In particular, some SSDs report incorrect temperatures attributes to the controller that cause chassis fan and system thermal issues. In addition, any SATA device used on that controller will be downclocked to 3Gbps speeds from 6Gbps. So you're ...


3

Well, doesn't the SSD choice depend a bit on your anticipated workload? Really: Are SSD drives as reliable as mechanical drives (2013)? But generically, you can attach just about anything to an LSI (Perc) controller and make it work. Should you? I mean, these are still consumer disks... There's no TRIM support on the hardware RAID controller (it's not ...


3

Here are the data sheets for the two drives: Intel S3500 LiteOn basically the LiteOn drive is slightly less performant, lower numbers in IOPS and Read/Write. Also I don't see write lifetime for the LiteOn, but the Intel is rated at 275TBW. Generally the Intel looks like a slightly better drive. That said, if you where expecting Intels and got Liteons ...


2

Buy the SSD drive that matches your expected workload. From the Dell SSD FAQ: The use of an endurance management algorithm ensures that sufficient Program/Erase (P/E) cycles are available for the warranty time period of the drive. The firmware will limit writes if a drive is written heavily. If you buy the a drive not intended for a write heavy ...


2

Depends on what you're doing to the SSDs. Today, MLC versus SLC doesn't matter as much as buying the appropriate drives with the right attributes for your use case and environment. If that means "write heavy", "read optimized", "low latency", whatever... See: Are SSD drives as reliable as mechanical drives (2013)? But why complicate your thinking? If a ...


2

Software RAID under Linux on modern hardware is fine... even with SSDs. It doesn't place a tremendous demand on your CPUs. Really. Heck, with premium Fusion-io solid-state drives, the recommended and common deployment scheme is to use software RAID. I wouldn't worry about this at all. Also see: Do I need to RAID Fusion-io cards?


2

I'll play devil's advocate here. If you are considering name-brand ENTERPRISE GRADE drives, with appropriate specifications, then in general they will be fine in commodity x64 servers. If you are considering CONSUMER GRADE drives, you are taking your chances. Other posters and commenters have explained the difference in quality and performance ...


2

16 disks (especially SSDs) in RAID5 is a bad idea... The HP P420i Smart Array controller would have warned you about this when you created the Logical Drive. There's some finesse needed to tuning the server you're describing. Can you provide specifics on the SSDs you're using, the current controller settings, the host's firmware levels and your testing ...


2

Your analysis is pretty correct. Use a few HDDs for lots of GBs, and lots of HDDs to a few IOps. Use a few SSDs for lots of IOPs, and lots of SSDs for a few GBs Which is more important for you? Space is the big cost-driver for SSD solutions, since the price-per-GB is much higher. If you're talking about a 200GB database needed 4K IOPs, a pair of SSDs will ...


2

Unless you bought a jank-tastic model that's been unreasonably stripped down, you only need the tray. The drive slides into the SATA ports on the backplane as its guided by the grooves that the tray fits in. I have tried surfing around but maybe it is so obvious that it is not even stated? I wasn't going to say anything, but since you brought the topic ...


1

You can't mix disk types in the same RAID group. However, you can mix drive types in the same server/controller. As for caching policies, they can be configured per drive group.


1

Please provide numbers detailing your expected and actual performance figures. Also, what is the SAS topology? How many SFF-8088 cables are in place between the host and the D2700 JBOD? As I mentioned earlier, the HP StorageWorks D2700 is S.M.A.R.T. aware and reports on SCSI Enclosure Services (SES) details... But your use case here is narrow. That's a ...


1

Bad news about MTBF is that common evaluation metodics suppose evenly distributed write load among all NAND cells. But cells are grouped into the clusters and when one single cell fails - whole cluster is marked as dead and is replaced with new one from the reserve. Usually reserve is about 20% of the SSD volume. When reserve is exhausted whole SSD will be ...


1

Assuming you keep it under warranty, if it ever fails, they'll replace it. The expected lifespan is their problem.


1

I don't know any good tools, directly from the server. Instead, I'd recommend answering their question with a question- if you're doing IO to SSD, which has no moving parts and thus does sequential IO at the same speed as random IO, why do they care? Are they trying to size new storage for you that would be partly on regular disk?



Only top voted, non community-wiki answers of a minimum length are eligible