Posts by Chris Mellor 1 357 publicly visible posts • joined 10 Jun 2009
Sloppy writing by me. Should have said pure 16TB drive. It can make the HC550 a 16TB drive by dropping a platter but WD guy said it was skipping pure 16TB drive level to go straight to 18TB and 20TB (with shingling). That's because it reckins it will be faster to market with 18 and 20TB drives than Seagate.
John C. Tran of HPE Global Product Communications tells us by mail: “Although Eyal David attempted to define our solution, it is factually incorrect. HPE Synergy, with built-in HPE OneView, is based on an open ecosystem with over 30 partners who support direct integrations, including VMware, Microsoft, Ansible by Red Hat, and Chef. HPE OneView was designed with an open REST API from its inception, and our partners provide key integrations to use HPE OneView directly with their products.
“HPE also provides popular open source SDKs and language bindings for PowerShell, Python Go, Java, Redfish, and Ruby. These direct integrations and tools allow customers to integrate with a wide range of data center orchestration platforms.”
Apparently NetAp's open source Trident container storage provisioner and orchestrator did PSO-type things for NetApp ONTAP and SolidFire storage back in 2016, as SPGoetze mentions above. I was pointed to this information - https://netapp.io/2016/12/23/introducing-trident-dynamic-persistent-volume-provisioner-kubernetes/ - by Gabriel Chapman via Twitter.
A vendor contact sent me this:
I think you might have made a mistake in your conclusion in this article.
In the last paragraph you say:
The destination storage array will provide data services by using "a global file system that handles replication/data protection at the file-system layer (HDFS, MongoDB, Casandra, NFS, etc.)". Comment from storage vendors will be most welcome. ®
I think if you re-read Greg’s comment, he talks about running global file systems outside of an array as an alternative to putting data services inside an array behind an NVMe namespace. So, he is suggesting you will either do it one way or another – without an array and using distributed software on hosts (i.e. like Excelero, or the apps he lists above), or the services will be contained within an array, like what Pavilion (and the NVMe-over-FC products down the road) does.
A little baffled by your comment "It's not likely you'll find much expertise around these days in digitising film reels" in the news roundup this week. How do you think older films make it into broadcast, shiny disc and streaming services?
The market for motion picture film scanning is robust, with companies like Arri, Digital Vision, DFT, Filmlight, MWA-Nova, Blackmagic and Lasergraphics all currently making high-end (4K and upwards) scanners. There are a lot of companies out there specialising in scanning and remastering motion picture movies, especially as studios realise how much money is to be made from their back-catalogues. Even us paupers in Auntie Beeb still do it. :D
Sent to me by a financial analyst:-
3D XPoint: Intel's Navin Shemoy misspoke on 14 November. The demo was not 100% in 3TB of 3D Xpoint but was impressive nonetheless. The Oracle World demo was in the last 10 minutes of Doug Fisher's presentation. Demo compared execution time for 3 DB queries between a server with 1.5TB of DRAM interacting with SSDs against that server with 1.5TB of DRAM interacting with 1.5TB of 3D XPoint DIMMs. The latter was roughly 8x faster. All three queries done markedly faster with the 3D XPoint DIMMs. Oracle said it has been working with Intel on the technology for four years. If interested, watch the last 10 minutes of https://www.oracle.com/openworld/on-demand.html?bcid=5595259086001
I hae learned from a correspondent who must remain anonymous that Brocade design their own ASICs for Fibre Channel switching. The actual chips may have been fabbed by IBM.
Brocade reckoned that having its own custom ASICs enables lower latency, more sophisticated monitoring functions, better power/cooling requirements as well as more reliable kit.
My correspondent suggests that we might think it strange that Cisco is seeking protection when historically it's played catch up with every generation of Brocade kit whilst trying to strangle FC as a technology.
Here;s a note I was sent by an Excelero person;
I want to point out that in contrast to E8, Excelero believes that sharing volumes is critical and therefore our volumes support such functionality providing distributed data-center grade data protection while being shared among multiple compute servers. We find it unfortunate that Pavilion chose to claim our solution provides JBOF only functionality, perhaps out of ignorance.
Pavilion are taking the path of gradual improvements to proprietary hardware appliances. This approach is outdated for today's modern data center whose architect are focusing on reducing overall TCO per unit of computation, storage and networking. Standard hardware combined with intelligent distributed scale-out software solutions are all trending up as they are a means to achieve these TCOs. Excelero NVMesh is the storage software of choice in this domain.
- - - - - - - -
That's it.
Chris.
I tried to get more data from Pure Storage and failed; customer not ready to talk. Can't see a like for like substitution here. Resigned myself to waiting for more info to come while thinking/judging that there was something real here if Pre Storage tweeters were getting that excited about it, even allowing for in-house Kool-Aid drinking. Hope that was a good call.
[Entered on behalf of Nimble employee...]
I am a Nimble employee for full disclosure.
Nimble has had variable block from day 1 on our hybrids (and now on All-flash which is the same OS) which came out a number of years prior to Pure releasing their product.
The below statement about our product is 100% false:-
"Side note: Nimble’s data reduction, for example, operates on fixed block sizes that have to be set/tuned on a volume-by-volume basis….in our minds that’s simply not correctly architected for today’s cloud/mixed workload reality."
On a second (and more political) note: Pure proves in their own diagram that very little IO actually happens at 32k... Transactional apps operate at 4-8k blocks and sequential apps happen at larger 64-256k blocks meaning 2 benchmark ranges are representative of real world workloads.
It's intellectually insulting for Pure to insist after showing their numbers that the mean of these is a valid benchmark. I would hope that such a fantastic team of analysts and journalists would call this faulty logic out.
[Obviously not :-) ]
This comment was posted privately to me, to spare my blushes maybe, but I figured it deserved a public posting anyway:-
"2.7e9 is greater than 2.65e9?
I would like to see another significant figure in that revenue report of $2.7 billion if it is being claimed that it exceeds a prediction of 2.65 billion.
So, a closer look, since I am a little bored and this might be a quick practice problem-
Seagate's Q4 supplemental report shows disk revenues of 2.455 billion (possibly erroneously reported as $2.455m in the article [ since corrected; thanks.] ) plus Flash/other of $0.199 billion, for a total of $2.654 billion. This matches the details of the Q4 report even though their own press release says $2.7 billion.
Technically 2.654 can be reported as 2.7, but I think maybe the opening paragraph is a little misleading? Or optimistic?"
Neat, huh?
This comment was sent privately to me, to spare my blushes maybe, but I figured it deserved a public posting - so I could get a public pasting :-)
"2.7e9 is greater than 2.65e9?
I would like to see another significant figure in that revenue report of $2.7 billion if it is being claimed that it exceeds a prediction of 2.65 billion.
So, a closer look, since I am a little bored and this might be a quick practice problem-
Seagate's Q4 supplemental report shows disk revenues of 2.455 billion (possibly erroneously reported as $2.455m in the article - [Corrected - thank you!] ) plus Flash/other of 0.199 billion, for a total of 2.654 billion. This matches the details of the Q4 report even though their own press release says 2.7 billion.
Technically 2.654 can be reported as 2.7, but I think maybe the opening paragraph is a little misleading? Or optimistic?"
Neat, huh?
I'm (Chris Mellor) posting this comment for Peter Serocka -
As poster Yaron Haviv pointed out, blades and nodes apparently are mixed up here, with most of the numbers relating to the 4U chassis-nodes, while "400+" relating to blades which will act as the functional nodes of OneFS in the sense how "nodes" are usually operated and counted in OneFS.
NAS Ops rates are least well-defined, so let's have a look at the other figures first, and see what we can figure out from the published information.
8 x 40GbitE frontend + 8 x 40GbitE backend sum up to 16 connectors per "node" and that must be
"chassis" as it would be overkill for any kind of sub-chassis storage "blades" to have 16 connectors on them. Just imaging the cabling...
One the other hand, present OneFS nodes have 2 x Infiniband backend plus 2 x 10GbitE frontend,
both active-active balancing and failover, so a Nitro blade should have at least 2 x 40GbitE
frontend and 2 x 40GbitE backend.
That would mean 4 blades per node.
(Otherwise, 4 x 40GbitE plus 4 x 40GbitE per blade would mean only 2 blades per chassis - barely makes sense, and wouldn't match the performance and capacity figures either, as we'll see below).
So assuming 4 blades per chassis, and that all capital "B"s mean Bytes not bits, and a max cluster size of 400 blades (aka "nodes" in OneFS speak), we might have:
1 Nitro Chassis (4U):
60 x 15 TB = 800 TB
15 GB/s (10 times more throughput ***out of 4U***, compared to one Isilon X410 4U node)
1 Blade:
15 x 15 TB = 225 TB
3.75 GB/s (2.5 times more throughput ***out of a single node*** in traditional sense, compared to X410).
Large Cluster (100 chassis = 400 blades):
100 x 800 TB = 400 x 225 TB = 80 PB (with 100 PB for "400+" blades)
100 x 15 GB/s = 1.5 TB/s (as claimed)
Note that 3.75 GB/s per blade fits nicely with the assumed network connections of dual redundant 40 GbitE for front and backend, respectively.
The picture of the cluster on Chad Sakac's blog site ***shows*** 100 chassis and ***says*** 400+ nodes, another indication that 1 "OneFS node" = 1 Nitro blade, with 4 of them in going in one Nitro chassis.
Fwiw, if we divide a chassis-node's 15GB/s by claimed 250,000 NAS Ops/s that would give
us an *average* NAS operation block size of 60 KB, which kind of makes sense for mixed
workloads (reads/writes at 100+ KB/transfer, and numerous namespace ops with at a few KB/transfer.) Same result when looking a a single blade of course, as dividing both throughput and Ops rates by 4 with cancel out.
With a latency of claimed 1 ms, the NAS queue depth per blade (= OneFS NAS node) would be 1/4 * 250,000/s x 0.001s = 62.5, which also is a reasonable value.
Makes sense?
---------------------
He adds this point: As for the confusion between "nodes" and "nodes":
With the Isilon OneFS software it is very clear what a NODE is: one instance of the OneFS FreeBSD-based operating system, running on a single SMP machine with disks enclosed.
But for the hardware guys "nodes" are those solid pieces of metal that get mounted in racks.
I think with Isilon clusters, one should keep the original OneFS definition of a node, and refer to the new hardware units in a different way. "Brick" hasn't been used with Isilon yet ;-)
Too bad EMC didn't sort out their terminology before making the Nitro pre-announcement, but such confusion arises often with bladed compute clusters, too.
Here's an update to Chad Sakac's blog post; "UPDATE: many people have scratched their heads at this – note that nowhere have I explicitly stated the relationships between blades and nodes (how many blades/node). That’s intentional. Lots of time before GA, and through that time, more will become evident. It’s not uncommon for some details to be left blank (sometimes to keep cards close to one’s chest, sometimes because there’s still variations likely in the plan). With Project Nitro we’re keeping some blade details back. In similar pre-GA statements from EMC and from almost everyone, there are some details kept back."
If a node contains more than one blade then working out the numbers could become easier.
In reply to Dr Bandwidth,
New table uploaded for your minor point. The major points relate to a chart created by Jim Handy of Objective Analysis (http://objective-analysis.com) and meant, in my understanding, as an indicative general table showing the relative values of various memory and storage media in the 2-dimensional space defined by the two axes. I'd suggest you take up the detail questions you have with Jim.Handy (at) Objective-Analysis.com.
Cheers ... Chris.
Hi Anonymous C .... The table is a much modified version of a source table some one gave me.I would dearly love more accurate numbers.
Re disk seek;. a Toshiba L200 (2.5-inch 5,200rpm) has a 5.56ms seek time so a 10ms seek time as an indicative number for disks doesn't seem that far out. My understanding is that 15,000rpm drives could have 4-5ms seeks, desktop 3.5-inch drives have 9-11ms seek times, and sluggish mobile 2.5-inchers could be as slow as 12ms. Do you have better numbers?
What would you say would be a better number for DAS access, assuming, say, a 10K rpm SATA drive?
Ditto SAN access?
Cheers .... Chris
Thanks Vaughn. That's me taken out to the woodshed - as I'm bush league :-)
So there are 20 drives in the coming top-end FlashArray//m-whatever controller enclosure. That would be extra 20 x 8TB = 160TB totalling up to (384 + 160) = 544TB, which makers 1.l5PB usable more achievable, as you say. Lovely system. Also the chart then needs updating.... sigh .... as the new FlashArray would be ahead of the VNXe3200 (500TB) but still behind Kaminario's K2 and its 740TB.
Thanks for this. I've sent your revised Pure numbers to Stifel's Aaron Rakers who provided the Pure revenue numbers you disagreed with. Let's see how he responds.
I've run your numbers through my personal spreadsheet and they do as you suggest, depressing Pure's line on the chart. It's still ahead of NetApp and HPE 3PAR but not by as much as before.
Chris.
Sent to me by a reader and posted anonymously:
Skyera at $400M? Not what I'm hearing on the street. They ran out of money, many leaving, forced to sell. I suspect anything close to $400M would have been material enough to require WD to disclose the actual amount as a public company. Since they were a key investor the amount was quite low.
-----------------
I heard it was a low amount from another person too.
Chris.
Sent to me and posted here anonymised:
you can tell the difference, like everybody can.
Take your favourite CD, that does not contain metallic rock or other music with a lot of white noise.
More like acoustic instruments, singing etc, without a lot of cymbals and other white noise producing instruments, will do perfectly
rip a MP 3 of that CD
play the CD and the MP3 player simultaneously, so you can alternate between the two, and hear the difference if any.
You *will* hear the difference, on a somewhat reasonable HiFi installation.
(Mailed to me:):
In your article on lossy vs. lossless audio you said, "Everything between sample points is lost." Please read up on the Nyquist–Shannon sampling theorem at Wikipedia, which states in part, "no actual 'information' is lost during the sampling process," given certain sampling conditions. This is scientific truth.
Please read the very long and detailed web page at https://xiph.org/~xiphmont/demo/neil-young.html for more information on lossy audio reproduction.
If you've heard of the placebo effect you will understand why people believe lossless audio reproduction MUST be better than lossy reproduction, and hear it as such. But the Canadian Research Council have conducted extensive double-blind (very important) listening tests. At sampling rates 256 Kbps and above, with a good encoding, only golden-ears individuals (that's not you or me or most people) can hear ANY difference at all.
(Name withheld)
Cheers .... Chris
There are a few other differences between the SSD Pro 1500 Series and Pro 2500 Series.
- The Pro 2500 Series is available in the channel (in the 2.5” form factor), while the Pro 1500 Series was not.
- The Pro 2500 Series also includes the Trusted Computing Group’s Opal 2.0 standard (Pro 1500 Series used the 1.0 standard), as well as Microsoft eDrive. -
Lastly, the Pro 2500 Series utilizes a second source of NAND, from SK Hynix. This SSD is the first from Intel to use a second source of NAND.
Chris.
Intel SSD Pro 2500 Series supports TCG’s latest Opal 2.0 features and is a Microsoft eDrive ready solution while the Intel SSD Pro 1500 supports Opal 1.0 features and does not support Microsoft eDrive. The Intel SSD Pro 2500 Series also supports more Advanced Power Management (APM) with 5 power states and 2 thermal states.
Chris.
Anand Tech editor Kristian Vatto says Pro 2500 differs from Pro 15000 by "TCG Opal 2.0 & eDrive compliant plus the Pro 2500 will be available in the channel as well (1500 was OEM only)"
PIty Intel didn't say that upfront. And, anyway, why not add these retrospectively to Pro 1500?
Chris.
Sent to me and posted here. I;m trying to find out from Storsimple what the actual HW configs are:-
I think those disk capacities are wrong.
The small unit has 10 bays and therefore 5 disk mirrors (assuming your analysis holds). 15TB would be 3TB disks, then. If it's 300GB drives, 1.5TB total storage. Either way - a correction needed I suspect.
The larger unit potentially holds 20 x 4TB drives (10 mirrors @ 40TB) plus SSD.
Chris.
This was mailed to me,
Chris------> I was at NASA Ames' N258 building (virtual wind tunnel) in 1997 when MegaDrive won the bake off against Clarrion and a couple of others. Not too bad- rebranded LSI Logic arrays- but had some bugs in the backplane that caused data corruption. Still, for the time, decent enough.
Decent guys all in all. DDN, which they became, also decent guys, albeit competitors when I was at Xyratex a couple of years ago.
(Name with held)
Here's more info from CTERA:
I should note that we support more than what is immediately available on our website… we need to update that :)
Amazon S3, Amplidata, Caringo, Cleversafe, Cloudian, DDN WOS, EMC Atmos, EMC ViPR, Hitachi HCP, IBM GPFS, OpenStack SWIFT, Scality.
With more to follow…
Additionally ... we have our own sync & share client that co-exists with our backup client so customers can manage backup and sync from one central system where the cloud is optionally extended by gateways (that also feature NAS support) - gateways are used when a customer wants LAN-level performance and low latency of data access.
BTW – you might have seen this one this week; Orange Business Services Launches CTERA-based Flexible Cloud Storage.
With 8M business customers at Orange, this is a tremendous validation of the platform.
(All this comes via mail from Jeff Denworth.)
Chris.
Well, well, well; a couple of Twitter people have pointed out that Actifio and IBM are connected; "The Actifio Protection and Availability Storage (PAS) appliance includes SVC code.[8] The PAS platform spans backup, disaster recovery, and business continuity among other functions." (Wikipedia.)
This webpage, dated Feb 2012, (http://mspmentor.net/managed-storage-services/actifio-ibm-partner-virtualized-storage-target-msps) is entitled "Actifio, IBM Partner On Virtualized Storage, Target MSPs." It says Actifio is producing a turnkey offering in partnership with IBM.
Data here - http://www.actifio.com/products/product-line/ - on Actifio 100T product line.
Chris.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
