QNAP Introduces Cheapest 10GbE NAS

[Julio Urquidi]

QNAP has announced its most affordable 10GbE NAS yet. The four bay TS-431X joins the 5-bay TS-531X and 8-bay TS-831X and has one 10 Gigabit Ethernet SFP+ port in addition to two Gigabit Ethernet ports.

The 10 GbE port allows single clients to take full advantage of its high transfer speeds. QNAP specs Windows large file tranfer reads at over 950 MB/s and writes at 525 MB/s. AES 256-bit encrypted performance is spec'd at 312 MB/s read and 240 MB/s write.

The TS-431X runs on Annapurna Labs’ Alpine AL-212 Dual-core 1.7 GHz ARM Cortex-A15 processor with 2 GB or 8 GB of DDR3 RAM. The TS-531X and TS-831X run on quad-core Annapurna Labs Alping AL-314 platforms.

The TS-431X’s hot swappable drive bays support four 3.5”/2.5” SATA HDDs/SSDs in RAID 0/1/5/6/10 and JBOD configurations. Three USB 3.0 ports support attached drives and printers.

The QNAP TS-431X will be available shortly at MSRPs of $349 for the 2 GB model and $499 for the 8 GB.

 

[RMerlin]

Seems to be an increasing amount of devices based on Alpine CPUs lately. I must say I find it a bit hard to believe that a 1.7 GHz ARM CPU would be able to push 950 MB/s of SMB throughput.

 

[sfx2000]

Seems to be an increasing amount of devices based on Alpine CPUs lately. I must say I find it a bit hard to believe that a 1.7 GHz ARM CPU would be able to push 950 MB/s of SMB throughput.

It does... comes down to Big ARM cores and the fabric inside the SoC itself...

The AL-212 is pretty nice - and Marvell's response is even better...

It's an odd place - and we see Intel throwing in with C3xx (Denverton) and Xeon-D bracketing that place...

Getting into bigger ARM's - Qualcomm with Centriq and MACon with X-Gene 3 - we shouldn't rule out Huawei's HiSilicon folks...

 

[hggomes]

950MB/s is simply impossible from usual mechanical HDD's RAID, but from SSD RAID0 it's not impossible, I'm able to squeeze similar throughput from a Quad-Core Intel J3160 RAID0.

 

[thiggins]

950MB/s is simply impossible from usual mechanical HDD's RAID, but from SSD RAID0 it's not impossible, I'm able to squeeze similar throughput from a Quad-Core Intel J3160 RAID0.

I agree with that. QNAP provides no details behind how the throughput was measured. But I think it's a safe assumption, SSD drives and/or SSD cache is involved.

We'll see soon enough. I'm getting one in for review.

 

[hggomes]

Maybe you can also try "openssl speed aes-256-cbc" results on Alpine CPU (with and without AES-NI) so I can make a compare with Intel J3160.


Thank you.

 

[sfx2000]

Maybe you can also try "openssl speed aes-256-cbc" results on Alpine CPU (with and without AES-NI) so I can make a compare with Intel J3160.

Communications Processors do tend to have some emphasis on Crypto, and with the right code, they can be very fast...

Armada 38x is one such beast, the Alpine is another - and then we have Intel with Rangley and Xeon-D (early numbers are suggesting that Denverton is going to be fairly fast here as well)

For testing, the command one should run...

openssl engine && openssl speed aes-256-cbc && openssl speed -evp aes-256-cbc​

Let's detect what engines are available...

(cryptodev) BSD cryptodev engine
(rsax) RSAX engine support
(rdrand) Intel RDRAND engine
(dynamic) Dynamic engine loading support

This is stock, without acceleration

Doing aes-256 cbc for 3s on 16 size blocks: 3975200 aes-256 cbc's in 2.98s
Doing aes-256 cbc for 3s on 64 size blocks: 1124177 aes-256 cbc's in 2.98s
Doing aes-256 cbc for 3s on 256 size blocks: 289151 aes-256 cbc's in 2.97s
Doing aes-256 cbc for 3s on 1024 size blocks: 188048 aes-256 cbc's in 2.98s
Doing aes-256 cbc for 3s on 8192 size blocks: 23848 aes-256 cbc's in 2.98s
OpenSSL 1.0.1s-freebsd  1 Mar 2016
built on: date not available
options:bn(64,64) rc4(16x,int) des(idx,cisc,16,int) aes(partial) idea(int) blowfish(idx) 
compiler: clang

The 'numbers' are in 1000s of bytes per second processed.
type             16 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes
aes-256 cbc      21312.07k    24108.01k    24933.95k    64523.11k    65461.89k

This is the -evp output, which is leveraging into the C2358's capabilities - not even QuickAssist, as that's not enabled yet

Doing aes-256-cbc for 3s on 16 size blocks: 704669 aes-256-cbc's in 0.26s
Doing aes-256-cbc for 3s on 64 size blocks: 664222 aes-256-cbc's in 0.24s
Doing aes-256-cbc for 3s on 256 size blocks: 548255 aes-256-cbc's in 0.27s
Doing aes-256-cbc for 3s on 1024 size blocks: 325671 aes-256-cbc's in 0.14s
Doing aes-256-cbc for 3s on 8192 size blocks: 67313 aes-256-cbc's in 0.05s
OpenSSL 1.0.1s-freebsd  1 Mar 2016
built on: date not available
options:bn(64,64) rc4(16x,int) des(idx,cisc,16,int) aes(partial) idea(int) blowfish(idx) 
compiler: clang

The 'numbers' are in 1000s of bytes per second processed.
type             16 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes
aes-256-cbc      43732.19k   175526.02k   513292.00k  2371463.85k 11763799.38k

This is on pfSense (FreeBSD) 2.3.3_1...

 

[sfx2000]

Maybe you can also try "openssl speed aes-256-cbc" results on Alpine CPU (with and without AES-NI) so I can make a compare with Intel J3160.

J3160 should be around 400K with AES I would expect - it's not too much different than N3700 for single threaded performance...

 

[hggomes]

From what i can remember it's 600K, but i can give it a try once again...

 

[sfx2000]

The 10 GbE port allows single clients to take full advantage of its high transfer speeds. QNAP specs Windows large file tranfer reads at over 950 MB/s and writes at 525 MB/s. AES 256-bit encrypted performance is spec'd at 312 MB/s read and 240 MB/s write.

I wonder if they mean little "b" and not big "B"....

 

[hggomes]

Talking about Alpine CPU's, we will see it soon available on a ASUS router model.


type 16 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
aes-256-cbc 43732.19k 175526.02k 513292.00k 2371463.85k 11763799.38k

I'm quite surprised on those results from a Atom C2358 Dual Core/2 Threads 1.7GHZ CPU (2013), Atom J3160 CPU it's a better one on single and multithread, so it should perform better.

 

[sfx2000]

Talking about Alpine CPU's, we will see it soon available on a ASUS router model.

The Alpine is a good chip - Amazon has been using this in-house for their clouds...