Synology DS2413+

Based on the recommendations of many members of the vExpert community, I purchased a Synology DS2413+. This is a 12-bay, Linux-based array that can be expanded to 24 spindles with the addition of the DX1211 expansion chassis. My plan was to eliminate a pair of arrays in my home setup (an aging Drobo Pro and my older iomega px6-300d), keeping a second array for redundancy.

The array is a roughly cube-shaped box which sits nicely on a desk, with easy access to the 12 drive trays and "blinky lights" on the front panel. It also sports two gigabit (2x1000Mb/s) network ports that can be bonded (LACP is an option if the upstream switch supports it) for additional throughput.

Synology has a page full of marketing information if you want more details about the product. The intent of this post is to provide the benchmark information for comparison to other arrays, as well as information about the device's comparative performance in different configurations.

The Synology array line is based on their "DSM" (DiskStation Manager) operating system, and as of this iteration (4.1-2661), there are several different ways to configure a given system. The result is a variety of different potential performance characteristics for a VMware environment, depending on the number of spindles working together along with the configuration of those spindles in the chassis.

The two major classes of connectivity for VMware are represented in DSM: You can choose a mix of NFS and/or iSCSI. In order to present either type of storage to a host, disks in the unit must be assembled into volumes and/or LUNs, which are in turn published via shares (NFS) or targets (iSCSI).

DSM supports a panoply of array types—Single-disk, JBOD, RAID0, RAID1, RAID5, RAID6, RAID1+0—as the basis for creating storage pools. They also have a special "SHR" (Synology Hybrid RAID) which automatically provides for dynamic expansion of the storage capacity when an even number of drive sizes are present; both single-drive- and dual-drive-failure protection modes are available with SHR on the DS2413+.

When provisioning storage, you have essentially two starting options: do you completely dedicate a set of  disks to a volume/LUN ("Single volume on RAID"), or do you want to provision different portions of a set of disks to different volumes and/or LUNs ("Multiple volumes on RAID")?

iSCSI presents a different sort of twist to the scenario. DSM permits the admin to create both "Regular files" and "Block-level" LUNs for iSCSI. The former reside as sparse file on an existing volume, while the latter is done with a new partition on either dedicated disks (Single LUNs on RAID) or a pre-existing disk group (Multiple LUNs on RAID). The "Regular files" LUN is the only option that allows for "thin provisioning" and VMware VAAI support; the Single LUN option is documented as highest-performing.

For purposes of comparison, the only mode of operation for the iomega px6-300d (which I've written about several times on this blog) is like using "Multiple Volumes/LUNs on RAID" in the Synology, while the older iomega ix2-200d and ix4-200d models operate in the "Regular files" mode. So the DSM software is far more versatile than iomega's StorCenter implementations.

So that leaves a lot of dimensions for creating a test matrix:

• RAID level (which is also spindle-count sensitive)
• Volume/LUN type
• Protocol

DS2413+			1 block seq read (IOPS)	4K random read (IOPS)	4K random write (IOPS)	512K seq write MB/s	512K seq read MB/s
Protocol	RAID	Disks
iSCSI	none	1	16364	508	225	117.15	101.11
	RAID1	2	17440	717	300	116.19	116.91
	RAID1/0	4	17205	2210	629	115.27	107.75
		6	17899	936	925	43.75	151.94
	RAID5	3	17458	793	342	112.29	116.34
		4	18133	776	498	45.49	149.27
		5	17256	1501	400	115.15	116.12
		6	15768	951	159	60.41	114.08
	RAID0	2	17498	1373	740	116.44	116.22
		3	18191	1463	1382	50.01	151.83
		4	18132	771	767	52.41	151.05
		5	17692	897	837	56.01	114.35
		6	18010	1078	1014	50.87	151.47
	RAID6	6	17173	2563	870	114.06	116.37
Protocol	RAID	Disks	1 block seq read (IOPS)	4K random read (IOPS)	4K random write (IOPS)	512K seq write MB/s	512K seq read MB/s
NFS	none	1	16146	403	151	62.39	115.03
	RAID1	2	15998	625	138	63.82	96.83
	RAID1/0	4	15924	874	157	65.52	115.45
		6	16161	4371	754	65.87	229.52
	RAID5	3	16062	646	137	63.2	115.15
		4	16173	3103	612	65.19	114.76
		5	15718	1013	162	59.26	116.1
		6
	RAID0	2	15920	614	183	66.19	114.85
		3	15823	757	244	64.98	114.6
		4	16258	3769	1043	66.17	114.64
		5	16083	4228	1054	66.06	114.91
		6	16226	4793	1105	65.54	115.27
	RAID6	6	15915	1069	157	64.33	114.94

While this matrix isn't a complete set of the available permutations for this device, when I stick with the 6-disk variations that match the iomega I already have in the lab, I've been stunned by the high latency and otherwise shoddy performance of the iSCSI implementation. Further testing with additional spindles did not—counter to expectations—improve the situation.

I've discovered the Achilles' Heel of the Synology device line: regardless of their protestations to the contrary about iSCSI improvements, their implementation is still a non-starter for VMware environments.

I contacted support on the subject, and their recommendation was to create the dedicated iSCSI target volumes. Unfortunately, this also eliminates the ability to use VAAI-compatible iSCSI volumes, as well as sharing disk capacity for NFS/SMB volumes. For most use cases of these devices in VMware environments, that's not just putting lipstick on a pig: the px6 still beat the performance of a 12-disk RAID1/0 set using all of Synology's tuning recommendations.

NFS performance is comparable to the PX6, but as I've discovered in testing the iomega series, NFS is not as performant as iSCSI, so that's not saying much... What to do, what to do: this isn't a review unit that was free to acquire and free to return...

Update:
I've decided to build out the DS2413+ with 12x2TB drives, all 7200RPM Seagate ST2000DM001 drives in a RAID1/0, and use it as an NFS/SMB repository. With over 10TB of formatted capacity, I will use it for non-VMware storage (backups, ISOs/media, etc) and low-performance-requirement VMware workloads (logging, coredumps) and keep the px6-300d I was planning to retire.

I'll wait and see what improvements Synology can make with their iSCSI implementation, but in general, don't see using these boxes for anything but NFS-only implementations.

Update 2:
Although I was unsatisfied with the DS2413+, I had a use case for a new array to experiment with Synology's SSD caching, so I tried a DS1813+. Performance with SSD was improved over the non-hybrid variation, but iSCSI latency for most VMware workloads was still totally unacceptable. I also ran into data loss issues when using the NFS/VAAI in this configuration (although peers on Twitter responded with contrary results).

On a whim, I went to the extreme of removing all the spinning disk in the DS1813+ and replacing them with SSD.

Wow.

The iSCSI performance is still "underwhelming" when compared to what a "real" array could do with a set of 8 SATA SSDs, but for once, not only did it exceed the iSCSI performance of the px6-300d, but it was better than anything else in the lab. I could only afford to populate it with 256GB SSDs, so the capacity is considerably lower than an array full of 2TB drives, but the performance of a "Consumer AFA" makes me think positively about Synology once again.

Now I just need to wait for SSD prices to plummet...

ix2-200 FrankenDisk Comparison

Okay, so you're a bit of a whack-job gadget geek, you read my previous post about upgrading an iomega ix2-200 by replacing the stock drives with SSDs, and now you want to know what sort of benefits might be gained for such a stunt.

Look no further... Here are my results.

The first chart is meant to make it clear that the improvements in replacing the 5900RPM spinning disk with SSD is marginal compared with the performance of putting the SSDs into a system that can really take advantage of them. Yes, the improvement is there, but the small, 200MHz ARM processor with 256MB RAM in the ix2-200 has nothing on a 2GHz quad-core with 8GB (my home desktop). But if you're interested in this experiment, you already knew that: you want to see how the ix2 fared...

The 4K read/write tests are meant to benchmark the outer-envelope for typical filesystem use: your standard Windows desktop is using NTFS with 4K clusters, so all your I/O is based on these units, whether or not you're actually reading/writing something of a different size. By doing 100% random, any cache in the system is pretty much overwhelmed, giving the real distinction between the two drive types.

The big-chunk sequential reads and writes represent workload that reflects video streaming (sequential read) or backup (sequential writes). Like the 4K random tests, this sort of workload also gets little benefit from cache and reflects the real throughput of  the drives.

Naturally, the reads fared better than the writes, and it's interesting to see the distinctions between the unit's iSCSI versus CIFS performance. In summary, you can better-than-double the performance of the device through upgrading to SSD.

Drive Performance

UPDATE: more data from other arrays & configurations
In an earlier post I said I was building a table of performance data from my experimentation with my new iomega ix2-200 as well as other drive configurations for comparison. In addition to the table that follows, I'm also including a spreadsheet with the results:

The Corners	1 block seq read (IOPS)	4K random read (IOPS)	4K random write (IOPS)	512K seq write MB/s	512K seq read MB/s	Notes
local SSD RAID0	10400	2690	3391	63.9	350.6	2 x Kingston "SSD Now V-series" SNV425
ix2 SSD CIFS	3376	891	308	25.7	40.4	2 x Kingston "SSD Now V-series" SNV425
ix2 SSD iSCSI	4032	664	313	29.4	38.5	2 x Kingston "SSD Now V-series" SNV425
local 7200 RPM SATA RAID1	7242	167	357	94.3	98.1	2 x Western Digital WD1001FALS
ix4 7200RPM CIFS**	2283	133	138	32.5	39.4	4 x Hitachi H3D200-series; **jumbo frames enabled
ix2 7200RPM CIFS	2362	125	98	9.81	9.2	2 x Hitachi H3D200-series
ix2 7200RPM iSCSI	2425	123	104	9.35	9.64	2 x Hitachi H3D200-series
ix4 7200RPM iSCSI**	4687	117	122	37.4	40.8	4 x Hitachi H3D200-series; **jumbo frames enabled
ix4a stock CIFS	2705	112	113	24	27.8	4 x Seagate ST32000542AS
ix4 stock iSCSI	1768	109	96	34.5	41.7	4 x Seagate ST31000520AS
ix4a stock iSCSI*	408	107	89	24.2	27.2	4 x Seagate ST32000542AS; *3 switch "hops" with no storage optimization introduce additional latency
ix2 stock CIFS	2300	107	85	9.85	9.35	2 x Seagate ST31000542AS
ix2 stock iSCSI	2265	102	84	9.32	9.66	2 x Seagate ST31000542AS
ix4 stock CIFS	4407	81	81	32.1	37	4 x Seagate ST31000520AS
DROBO PRO (iSCSI)	1557	71	68	33.1	40.5	6 x Seagate ST31500341AS + 2 x Western Digital WD1001FALS; jumbo frames
DROBO USB	790	63	50	11.2	15.8	2 x Seagate ST31000333AS + 2 x Western Digital WD3200JD
DS2413+ 7200RPM RAID1/0 iSCSI	12173	182	194	63.53	17.36	2 x Hitachi HDS722020ALA330 + 6 x HDS723020BLA642
DS2413+ 7200RPM RAID1/0 NFS						2 x Hitachi HDS722020ALA330 + 6 x HDS723020BLA642
DS2413+ SSD RAID5 iSCSI	19238	1187	434	69.79	123.97	4 x Crucial M4

PX6-300			1 block seq read (IOPS)	4K random read (IOPS)	4K random write (IOPS)	512K seq write MB/s	512K seq read MB/s
Protocol	RAID	Disks
iSCSI	none	1	16364	508	225	117.15	101.11
	RAID1	2	17440	717	300	116.19	116.91
	RAID1/0	4	17205	2210	629	115.27	107.75
		6	17899	936	925	43.75	151.94
	RAID5	3	17458	793	342	112.29	116.34
		4	18133	776	498	45.49	149.27
		5	17256	1501	400	115.15	116.12
		6	18022	1941	1065	52.64	149.1
	RAID0	2	17498	1373	740	116.44	116.22
		3	18191	1463	1382	50.01	151.83
		4	18132	771	767	52.41	151.05
		5	17692	897	837	56.01	114.35
		6	18010	1078	1014	50.87	151.47
	RAID6	6	17173	2563	870	114.06	116.37
Protocol	RAID	Disks	1 block seq read (IOPS)	4K random read (IOPS)	4K random write (IOPS)	512K seq write MB/s	512K seq read MB/s
NFS	none	1	16146	403	151	62.39	115.03
	RAID1	2	15998	625	138	63.82	96.83
	RAID1/0	4	15924	874	157	65.52	115.45
		6	16161	4371	754	65.87	229.52
	RAID5	3	16062	646	137	63.2	115.15
		4	16173	3103	612	65.19	114.76
		5	15718	1013	162	59.26	116.1
		6	16161	1081	201	63.85	114.63
	RAID0	2	15920	614	183	66.19	114.85
		3	15823	757	244	64.98	114.6
		4	16258	3769	1043	66.17	114.64
		5	16083	4228	1054	66.06	114.91
		6	16226	4793	1105	65.54	115.27
	RAID6	6	15915	1069	157	64.33	114.94

About the data

After looking around the Internet for tools that can be used to benchmark drive performance, I settled on the venerable IOmeter. Anyone who has used it, however, knows that there is an almost infinite set of possibilities for configuring it for data collection. In originally researching storage benchmarks, I came across several posts that suggest IOmeter along with various sets of test parameters to run against your storage. Because I'm a big fan of VMware, and Chad Sakac of EMC is one of the respected names in the VMware ecosystem, I found his blog post to be a nice start when looking for IOmeter test parameters. His set is a good one, but requires some manual setup to get things going. Also in my research, I came across a company called Enterprise Strategy Group which not only does validation and research for hire, they've published their custom IOmeter workloads in an IOmeter "icf"configuration file. The data published above was collected using their workload against a 5GB iobw.tst buffer. While the table above represents "the corners" for the storage systems tested, I also captured the entire result set from the IOmeter runs and have published the spreadsheet for additional data if anyone is interested.

px6-300 Data Collection

The data in the px6-300 tables represents a bit of a shift in the methodology: the original data sets were collected using the Windows version of iometer, while the px6-300 data was collected using the VMware Labs ioAnalyzer 1.5 "Fling". Because it uses the virtual appliance, a little disclosure is due: the test unit is connected by a pair of LACP-active/active 1Gb/s links to a Cisco SG-300 switch. In turn, an ESXi 5.1 host is connected to the switch via 4x1Gb/s links, each of which has a vmkernel port bound to it. The stock ioAnalyzer's test disk (SCSI0:1) has been increased in size to 2GB and is using an eager-zeroed thick VMDK (for iSCSI). The test unit has all unnecessary protocols disabled and is on a storage VLAN shared by other storage systems in my lab network. The unit is otherwise idle of any workloads (including the mdadm synchronization that takes place when configuring different RAID levels for disks, a very time-consuming process); there may be other workloads on the ESXi host, but DRS is enabled for the host's cluster, and if CPU availability were ever an issue in an I/O test (it isn't), other workloads would be migrated away from the host to provide additional resource.

The Takeaway

As expected, the SSD-based systems were by-far the best-performing on a single-spindle basis. However, as one might expect, an aggregate of spindles can provide synergy that meets or exceeds the capability of SSD, and locally-attached storage can also make up the difference in I/O performance. The trade-off, of course, is the cost (both up-front and long-term) versus footprint.