Ultra 2 Plus with 2*WD Red 3GB disks --> Connected to a GS608 switch (A) Switch (A) is connected to another GS608 switch (B) Computers are connected to (B).
I feel my performance has degraded in the last 2 months, so I made some tests
Drag/Drop a 1.5GB file from NAS to machine connected via (B) = approx 3MB/sec Drag drop same file to same machine connected via (A) = approx 20MB/sec Drag drop same file to same machine connected with cable that connects (A) --> (B) = approx 20MB/sec Drag drop same file to same machine connected direct to 2nd port on NAS = approx 30MB/sec
In all cases, speed starts higher then drops to final speed.
I ran Bonnie on the server and got about 95MB/sec block sequential output.
I had noticed that the connections to (B) did not always show as 1Gb connections, so I bought another GS608 and replaced (B) with it.
Still same results.
Does having a second switch in the equation really mean I lose speed like that. Should I remove (A) and connect NAS direct to (B)? I've not tried that, but I'd like to have switch (A).
I have another NAS NV+, which is connected to (A) and is used for nightly backups. Could I connect that to the Ultra 2 Plus via the second LAN?
You shouldn't be losing speed with two switches in the path, unless there is a lot of other network traffic on the path between the switches. Did you try looking at the network stats on the NAS before and after the test? You'll find a "show errors' button on the network page.
I suggest you should do a test connecting the NAS directly to B. If that works, then it suggests something might be wrong with (A). I would also check the PC settings and make sure flow control is enabled.
Are the two switches in the same location?
BTW, you might also check the switch warranty status at support.netgear.com. Some have quite long warranties (lifetime for some).
I don't recommend trying to connect the NV+ via the second LAN port. One side effect is that you won't be able to access the NV+ admin page, unless you start messing around with the ultra-2's routing tables with ssh.
You are not the first to post this. Here's a note from a few years back. I recalled reading this when I was deciding what switch to move to next and google jumped right to it :-). Unfortunately this other guy determined a third non GS608 at the top of the switch hierarchy avoided the problem. Maybe you have a router you could try attaching the gs608's to and avoid cascading them?
Changing the link between A and B to links to your router is a reasonable idea if your router is gigabit. If not, you could try using the original "B" GS608 as dsm1212 suggests - connecting it to the router, and the other two switches to it.
BTW, you don't want loops, so A and B are both already connected to the router you'd need to either remove the link between them, or the link between one of them and the router.
Though they are more expensive, smart switches give you more options to diagnose these problems since they have their own web interfaces that show you port status and stats. Also, if "A" and "B" are co-located then you could get a 16 port switch. The "trunk" between the switches can only carry a gigabit in each direction. The internal switching fabric of most switches doesn't have that limitation.
Thanks for the useful replies. The port lights were green, unlike that post you mentioned. I am just using the 2 switches chained from the router. The router is Gb and has only a short cable to switch B, which is upstairs. I then have a 15m cable from B to the downstairs switch A, which is physically next to the NAS.
I connected the NAS directly to B with the 15m cable and all was good, so I then connected the upstairs end to the router and then accessed it through B, so via 2 hops, and all good there too.
So it looked like A was the problem as you suggested. I replaced A with the original upstairs switch, that I had replaced thinking it was the broken one, and now it gives me reasonable speed again.
I took apart A and the two capacitors were slightly bulging, which others had said had caused their problems
