Hot answers tagged

24

Assuming the graph is http request time it seems fairly reasonable to me. A http request (in the absense of keepalive, fastopen etc) normally requires at least two round trips. Client sends syn Sever receives syn and sends syn-ack Client receives syn-ack and sends ack and request. Server sends response. The speed of light in fiber is about 2*10^8 ...


18

Any other possible reason in addition to distance from server? The path the packets take.


8

Searching google for ping "us-mid" yields Monitis.com and their IP in Dallas, who also have a DE IP in Frankfurt. From a very well-connected server in France I have 9 ms ping RTT to the DE IP and 111 ms ping RTT to the US-MID IP. For HTTP time-to-response I'd expect double that plus the server's reaction time, for the sake of argument let's say maybe 26 and ...


6

Kevin mentions a number of great points, but it might help to start the analysis at a little higher level: what are the two machines and, since we're talking about Tomcat which runs in a JVM, do both have zAAPs or zIIPs (assuming zAAP on zIIP)? From the "d m=cpu" you should be able to get the machine model information which will at least let you know if you'...


3

Is there any impact on performance from a kvm guest if I create ext4 on this large drive directly, mount it up on the host like /mnt/kvms and then create raw kvm disk image on it? Every time, when your data have more layers to go through performance degradation and latency increase will appear. For example: Qemu RAW file -> Ext4 -> LVM -> RAID1 device -...


3

That's a pretty old RAID controller (those were introduced in 2010, I think?), and its throughput characteristics were designed for spinning disks, not SSDs. You might be bottlenecking on older gen SATA, or the RAID controller itself - but either way, that RAID controller isn't worth wasting your time on, performance-wise.


3

NFSv4.2 does have a offload-copy operation which can do server-to-server copy without proxying data trough the client. Modern linux kernels (> 3.13?) supports that. I don't know about other servers. UPDATE By linux kernel 4.7 server side copy is not supported by linux servers https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/fs/nfsd/...


3

First, it is perfectly normal that DM devices does not have any I/O scheduler, as (with specific exceptions) About the low performance you recorded, consider that your H310 controller not only has no cache, but it even disables the physical disk's DRAM cache, meaning your system has no way to lower latency via caching. Combining that with encryption, where ...


2

Random seek times will be roughly O(1) like uncompressed file systems as well, but with the caveat that up to 128 KiB of data is compressed together so to read just a single byte, all the data in that 128 KiB block will have to be read and decompressed. Depending on the access pattern, this can have a somewhat large performance impact, but you need to ...


2

The problem was https://developers.google.com/speed/pagespeed/module/ which have done cache clean every hour. Configuration it to store files on tmpfs fixed the problem.


2

/dev/sdc2 failed. It will need to be replaced.


2

c-plan-ahead have to set a positive value to enable dynamic sync rate controller. disk c-plan-ahead 15; // 5 * RTT / 0.1s unit,in my case is 15 c-fill-target 24; c-max-rate 720M;


1

This should be a comment, but its a bit long. with the database stopped It's a very unusual database which runs optimally on RAID 0 It's also somewhat unusual to spend so much money on a server and plan no resilliency for the storage. But to then compound that, by using raid 0 means you are three times more likely to loose all your data than with a ...


1

The "copy a file" operation is not a basic file-system operation such as read, write, open, close, and similar operations. See this page for a good explanation of the operations a filesystem has to support. And that's all file systems - whether it's ext4, btrfs, or nfsv4. Note that there's no way to know why a process opens file A, opens file B, reads ...


1

Nginx will always create new proxy connection for incoming connection unless it is cached localy. As MJPG is an infinite loop, it doesn't seem as an option here. What I would however try is setup a ffmpeg/vlc on the proxy server to consume mjpg stream from rpi. Nginx then can connect to local ffmpeg/vlc to get stream. You will then have only one connection ...


1

Scientific Method is your friend. You or someone on your behalf will have to monitor your system to gather the necessary data to allow you to make an informed decision. Your load average spikes quite sharply and at predictable times, that means it should be fairly straightforward to monitor the event with tools like top, iostat, vmstat etc and start the ...


1

Looks like you are on a Virtual server (VPS). If your VPS is based on OpenVZ, openVZ usually doesn't expose all counters (like /proc/diskstats) to its containers.


1

Is this an Internet facing server, or an Intranet facing server? I wouldn't mind having a look at it myself. Theoretically, there shouldn't be any reason why IIS is slow to serve static content - they're all routed through the static file handler, which is extremely efficient. Some things you can try: Ensure the websites never idle out (resulting the IIS ...



Only top voted, non community-wiki answers of a minimum length are eligible