Hi, I just recently got a DS1821+ i'm very new to Synolgy, I've been with Drobo for more than 10 years however since they went under I needed to find a new product that supported hybrid raid with a close match to a DAS setup as possible. I have set up ATTO ISCSI Initiator on my Mac and the Lun's on the Synology however I'm only getting about 110MBps read & write compared to my Drobo's 300MBps read & write. I do have SSD caching on there as well as the Synology.
Current hardware Setup for Synology:
32 gigs of ram
2TB NVME Cache X1 (read only), will put in a 2nd one soon for write as well.
Seagate 16TB Exo [ST16000NM001G] Enterprise Drives X3
Drobo has the same model HDD's as well.
now I have messed around with MTU with varying success, I either get faster write but read takes a big hit or vise versa..
I have also tested 2 network nodes instead of the 1 via the ISCSI Initiator. so Link Aggregation more less via 2 network cables & read & write does hit around 170MBs but is still very slow compared to Drobo's speeds.
I Have Stripe Cache size set to small for better IO
also, the storage pools are still optimising in the background not sure if this affects anything..
also, no data has been transferred yet as I want to get this IO issue resolved first..
also I ran iperf command
and no issues on the network side of things.
I'm also using Blackmagic Disk Speed Test to test the IO speed.
so I'm not sure what to do here, its like the SSD cache is not being utilised first for all new disk write/read operations..
is there anything i can tweak so it uses the cache first rather than writing to disk?
or do i need to wait till the storage pools are done optimising?
because with Drobo it uses the SSD cache first & when it fills up it then slows down IO till the SSD cache offloads to the main storage array, I would have assumed Synology does this as well?
Current hardware Setup for Synology:
32 gigs of ram
2TB NVME Cache X1 (read only), will put in a 2nd one soon for write as well.
Seagate 16TB Exo [ST16000NM001G] Enterprise Drives X3
Drobo has the same model HDD's as well.
now I have messed around with MTU with varying success, I either get faster write but read takes a big hit or vise versa..
I have also tested 2 network nodes instead of the 1 via the ISCSI Initiator. so Link Aggregation more less via 2 network cables & read & write does hit around 170MBs but is still very slow compared to Drobo's speeds.
I Have Stripe Cache size set to small for better IO
also, the storage pools are still optimising in the background not sure if this affects anything..
also, no data has been transferred yet as I want to get this IO issue resolved first..
also I ran iperf command
Code:
[ ID] Interval Transfer Bandwidth
[ 1] 0.00-10.00 sec 1.09 GBytes 940 Mbits/sec
and no issues on the network side of things.
I'm also using Blackmagic Disk Speed Test to test the IO speed.
so I'm not sure what to do here, its like the SSD cache is not being utilised first for all new disk write/read operations..
is there anything i can tweak so it uses the cache first rather than writing to disk?
or do i need to wait till the storage pools are done optimising?
because with Drobo it uses the SSD cache first & when it fills up it then slows down IO till the SSD cache offloads to the main storage array, I would have assumed Synology does this as well?