From mboxrd@z Thu Jan 1 00:00:00 1970 From: Benjamin ESTRABAUD Subject: Re: Best configuration for bcache/md cache or other cache using ssd Date: Thu, 19 Sep 2013 16:49:48 +0100 Message-ID: <523B1D1C.2070807@mpstor.com> References: <523A60DF.5040702@hardwarefreak.com> <523AAC1F.6050006@hardwarefreak.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: Sender: linux-raid-owner@vger.kernel.org To: Roberto Spadim Cc: Stan Hoeppner , Drew , Linux-RAID List-Id: linux-raid.ids On 19/09/13 16:30, Roberto Spadim wrote: > Hi Stan! Hi Roberto, Just a few things: > thanks a lot about your experience > I have some doubts about raid boards, what you look when you buy one? > i bought some raid boards (most perc from dell, others adaptec and > others lsi) and don't know if it's really a good feature... check if > i'm wrong about things i most look before buying one: > 1) Smart or other tool to diagnostics and access drives diagnostics > 2) Cache memory (if i have 512mb here, i could replace with 512mb or > more at linux side? instead of cache at raid board, why not add cache > to linux kernel?) You could, but your Linux box is unlikely battery backed. The advantage of these card's memory is that in the event of a power failure, the contents of memory is kept and flushed later on. This is especially important because here we are talking about "fast" writes: the RAID card tells the IO requester that the IO has been flushed even though it hasn't (at least not on the drive). If this was a Linux box and you lacked battery backing, the client would assume some data was written while it was actually lost, causing silent failure (the worst kind of it). > 3) batery backup, how this really work? what kind of raid board really > work nice with this? As Stan mentioned, some LSI controllers support this. You can lookup the LSI website to find out about it. Sometimes the BBU (battery unit) comes separately. > 4) support for news drivers (firmware updates) LSI is quite good when it comes to Linux drivers. They are updated directly in the official kernel so no need to update .ko files or anything like that. > 5) support for hot swap These cards usually support hot swap very well, with less downtime between pulling/pushing drives. (sometimes MD can take a few seconds to remove/add a drive in an array). > 6) if i use ssd what should i consider? i have one raid card with ssd > and i don't know if it's runs nice or just do the job > 7) anything else? costs =) ? The costs, as Stan mentioned also, are fairly reasonable, in the order of a good quality SSD drive. > > i will search about this boards you told, and about features (i don't > know what bbu means yet, but will check... any good raid boards > literarture to read? maybe wikipedia?) BBU means "battery backed unit" as far as I know. > > thanks a lot!! :) One note: This is not an LSI promotional response, I just know about LSI cards more than their counterpart, so I gave insight on what I knew. Regards, Ben. > > 2013/9/19 Stan Hoeppner : >> On 9/18/2013 10:42 PM, Roberto Spadim wrote: >>> nice, in other words, is better spend money with hardware raid cards >>> right? >> If it's my money, yes, absolutely. RAID BBWC will run circles around an >> SSD with a random write workload. The cycle time on DDR2 SDRAM is 10s >> of nanoseconds. Write latency on flash cells is 50-100 microseconds. >> Do the math. >> >> Random write apps such as transactional databases rarely, if ever, >> saturate the BBWC faster than it can flush and free pages, so the >> additional capacity of an SSD yields no benefit. Additionally, good >> RAID firmware will take some of the randomness out of the write pattern >> by flushing nearby LBA sectors in a single IO to the drives, increasing >> the effectiveness of TCQ/NCQ, thereby reducing seeks. This in essence >> increases the random IO throughput of the drives. >> >> In summary, yes, a good caching RAID controller w/BBU will yield vastly >> superior performance compared to SSD for most random write workloads, >> simply due to instantaneous ACK to fsync and friends. >> >>> any special card that i should look? >> If this R420 is the 4x3.5" model then the LSI 9260-4i is suitable. If >> it's the 8x2.5" drive model then the LSI 9260-8i is suitable. Both have >> 512MB of cache DRAM. In both cases you'd use the LSI00161/ LSIiBBU07 >> BBU for lower cost instead of the flash option. These two models have >> the lowest MSRP of the LSI RAID cards having both large cache and BBU >> support. >> >> In the 8x2.5" case you could also use the Dell PERC 710, which has built >> in FBWC. Probably more expensive than the LSI branded cards. All of >> Dell's RAID cards are rebranded LSI cards, or OEM produced by LSI for >> Dell with Dell branded firmware. I.e. it's the same product, same >> performance, just a different name on it. >> >> Adaptec also has decent RAID cards. The bottom end doesn't support BBU >> so steer clear of those, i.e. 6405e/6805e, etc. >> >> Don't use Areca, HighPoint, Promise, etc. They're simply not in the >> same league as the enterprise vendors above. If you have problems with >> optimizing their cards, drivers, firmware, etc for a specific workload, >> their support is simply non existent. You're on your own. >> >>> 2013/9/18 Stan Hoeppner : >>>> On 9/18/2013 12:33 PM, Roberto Spadim wrote: >>>>> Well the internet link here is 100mbps, i think the workload will be a >>>>> bit more than only 100 users, it's a second webserver+database server >>>>> He is trying to use a cheaper server with more disk performace, Brazil >>>>> costs are too high to allow a full ssd system or 15k rpm sas harddisks >>>>> For mariadb server i'm studing if the thread-pool scheduler will be >>>>> used instead of one thread per connection but "it's not my problem" >>>>> the final user will select what is better for database scheduler >>>>> In other words i think the work load will not be a simple web server >>>>> cms/blog, i don't know yet how it will work, it's a black/gray box to >>>>> me, today he have sata enterprise hdd 7200rpm at servers (dell server >>>>> r420 if i'm not wrong) and is studing if a ssd could help, that's my >>>>> 'job' (hobby) in this task >>>> Based on the information provided it sounds like the machine is seek >>>> bound. The simplest, and best, solution to this problem is simply >>>> installing a [B|F]BWC RAID card w/512KB cache. Synchronous writes are >>>> acked when committed to RAID cache instead of the platter. This will >>>> yield ~130,000 burst write TPS before hitting the spindles, or ~130,000 >>>> writes in flight. This is far more performance than you can achieve >>>> with a low end enterprise SSD, for about the same cost. It's fully >>>> transparent and performance is known and guaranteed, unlike the recent >>>> kernel based block IO caching hacks targeting SSDs as fast read/write >>>> buffers. >>>> >>>> You can use the onboard RAID firmware to create RAID1s or a RAID10, or >>>> you can expose each disk individually and use md/RAID while still >>>> benefiting from the write caching, though for only a handful of disks >>>> you're better off using the firmware RAID. Another advantage is that >>>> you can use parity RAID (controller firmware only) and avoid some of the >>>> RMW penalty, as the read blocks will be in controller cache. I.e. you >>>> can use three 7.2K disks, get the same capacity as a four disk RAID10, >>>> with equal read performance and nearly the same write performance. >>>> >>>> Write heavy DB workloads are a post child for hardware caching RAID devices. >>>> >>>> -- >>>> Stan >>>> >>>> >>>> >>>> >>>>> 2013/9/18 Drew : >>>>>> On Wed, Sep 18, 2013 at 8:51 AM, Roberto Spadim wrote: >>>>>>> Sorry guys, this time i don't have a full knowledge about the >>>>>>> workload, but from what he told me, he want fast writes with hdd but i >>>>>>> could check if small ssd devices could help >>>>>>> After install linux with raid1 i will install apache mariadb and php >>>>>>> at this machine, in other words it's a database and web server load, >>>>>>> but i don't know what size of app and database will run yet >>>>>>> >>>>>>> Btw, ssd with bcache or dm cache could help hdd (this must be >>>>>>> enterprise level) writes, right? >>>>>>> Any idea what the best method to test what kernel drive could give >>>>>>> superior performace? I'm thinking about install the bcache, and after >>>>>>> make a backup and install dm cache and check what's better, any other >>>>>>> idea? >>>>>> We still need to know what size datasets are going to be used. And >>>>>> also given it's a webserver, how big of a pipe does he have? >>>>>> >>>>>> Given a typical webserver in a colo w/ 10Mbps pipe, I think the >>>>>> suggested config is overkill. For a webserver the 7200 SATA's should >>>>>> be able to deliver enough data to keep apache happy. >>>>>> >>>>>> In the database side, depends on how intensive the workload is. I see >>>>>> a lot of webservers where the 7200's are just fine because the I/O >>>>>> demands from the database are low. Blog/CMS systems like wordpress >>>>>> will be harder on the database but again it depends on how heavy the >>>>>> access is to the server. How many visitors/hour does he expect to >>>>>> serve? >>>>>> >>>>>> >>>>>> -- >>>>>> Drew >>>>>> -- >>>>>> To unsubscribe from this list: send the line "unsubscribe linux-raid" in >>>>>> the body of a message to majordomo@vger.kernel.org >>>>>> More majordomo info at http://vger.kernel.org/majordomo-info.html >>>>> >>>>> >>> >>> > >