New answers tagged

0

8x 5400 RPM disks will generally outperform 4x 7200 RPM disks, save for single-threaded sequential read/write. As your use case (virtual machines) is rich of small random reads/writes, you should prefer the solution with more spindles (ie: the more the disks, the better). Anyway, considered how cheap are 7200 RPM disks, I strongly suggest to use 8x such ...


0

For a fool-proof strategy you can try proxmox, so you virtualize your dedicated server and can take advantage of kvm virtualization if planning to use Windows os, or lxc for recent linux distro. Then you can easily snaphot each VM into external ftp/nfs + some 'ol good file level backup (attic backup, acronis, rsync, rsnapshot, rdiff backup, bacula, and so ...


0

The server did not have enough memory. Instead of caching NTFS metafile data in memory every file access required multiple disk reads. As usual, the issue is obvious once you see it. Let me share what clouded my perspective: The server showed 2 GB memory available both in Task Manager and RamMap. So either Windows decided that the available memory was not ...


2

You would be better off using a tool like iperf instead of dd | nc. That way you can test both TCP and UDP and get a proper throughput report, as well as controlling the rate, packet size, etc. People regularly use this to measure 10Gbe and 40Gbe performance.


1

Watch top while running your test. It is possible you are hitting CPU limits on the nc command. We've seen nc become the limit to throughput by being CPU bound. Because we observe much higher throughput on kernel space things like DRBD replication, my guess is that nc experiences the overhead of many system calls. If this is the case, you can run ...


4

SATA/SAS are not like hubs, rather they work much like a network switch. This means that a single SATA/SAS port provides the entire advertised banwdwith, independetly from the other ports. In other words, a chipset providing 4x SATA 3.0 ports (6 Gb/s each) has a "SATA backplane" switching capability of 4x6 Gb/s = 24 Gb/s. For a 4x SAS 3.0 ports (12 Gb/s ...


0

Speed depends on the controller that powers the ports and the PCIe connection of the mainboard. In theory, given a sufficient controller chip and PCIe lanes from the CPU, each port would be independent and have maximum speed (except for about 10% protocol overhead), there are no shared resources like with PATA. In practice however, you get a wide range of ...


1

It is 6 Gb/s per port, with 8 ports you can get up to 48 Gb/s total. For SAS controllers and disks, you can use "SAS Expanders" to connect one SAS port to multiple drives. In this case, all drives connected to one SAS port share the total 6 Gb/s bandwidth. For this reason, for maximum performance we avoid SAS expanders and use controllers like the LSI ...


5

Rob-d mentions that load-balancers must perform health checks on the backend servers to ensure they're healthy and can serve requests. This is absolutely true and I think it's what would enable you to do what you want (checking other metrics and having the LB make routing choices based on those). Assuming you're load-balancing HTTP, most load-balancers will ...


3

The question you're asking is an extremely important question in relation to load balancing, there are Two primary reasons why we load balance, first and most obvious is to split client requests to 2 or more servers, second to make that service highly available. Having configured the load balancer with these two outcomes in mind we then enter into the realm ...


3

Typically the loadbalancing component of the loadbalancer is only only aware of either the number of active network connections and/or requests it has sent to a back-end server and knows nothing of the actual load those generate on the back-end system. The load balancing algorithm you select determines which back-end server will handle the next new ...


3

You are misreading your tools. Your one-minute-average system loads are 5.64 and 3.03 respectively (top line, right-hand end). The 74.9 and 41.2 figures are the percentage of time the CPU spends servicing system requests, usually IO or memory. Given that swap is being fairly heavily used, my guess would be memory. You can get a better feel for that by ...


0

TcpTimedWaitDelay is fully supported in Windows Server 2012 You may have to set the StrictTimeWaitSeqCheck as well, for TcpTimedWaitDelay to be of effect: [HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters] "StrictTimeWaitSeqCheck"=dword:00000001 Setting or changing these will require a reboot for the changes to be in effect. There is ...


10

Load is a very often misunderstood value on Linux. On Linux it is the measurement of all tasks in the running or uninterruptible sleep state. Note this is tasks, not processes. Threads are included in this value. Load is calculated by the kernel every five seconds and is a weighted average. That is the minute load is the average of 5/60, the five minute ...


-1

From my experience the general rule is directly related to how large your public key is (eg. 2048, vs 4096, vs 8192) all take significantly longer. However I can hardly notice a difference in a Desktop environment, but mobile is where you see a difference since it takes computing power. In general it is unfortunate but SSL has always and will likely always ...


4

From Wikipedia: However, Linux also includes processes in uninterruptible sleep states (usually waiting for disk activity), which can lead to markedly different results if many processes remain blocked in I/O due to a busy or stalled I/O system In other words, the load average reported by Linux includes any processes waiting for I/O (eg: disk or ...


0

Take note of the disk bottleneck : Disk IOPS. RDS will throttle IO request if it is not provisioned. There is 3iops/GB given to SSD. So if you have 100GB allocated for SSD, you can max up 3x100 = 300 iops block. If you pay for provisioned SSD, you can provision iops more than that. While for standard(magentic) storage, additional per millions iops charge ...


2

I'm going to turn my comments above into an answer. T2 instances only get a fraction of a CPU - 10% for a t2.micro, 40% for a t2.medium. You get CPU credits that build up, but once you use them you get throttled for CPU. You also get lower network and IO performance as you have to share physical machine resources. What I suspect is happening is your ...


2

Perhaps there was a time early on in the development of VT-x when it would reduce performance by a single percentage or two but those times are long gone and I wouldn't expect to see any impact whatsoever now. Obviously if you're using any form of virtualisation disabling it will have enormous negative impact but I'm assuming you know that. The reason it's ...


1

This kind of graphical analysis is usually done with tools that are designed to do so. I use munin, but there are several other tools that do similar tasks. They gather the data into a data store from which graphs can be readily generated. However, there are tools that can extract the data into formats that can be graphed. I would use awk, perl, or python ...


1

Google Pagespeed for Nginx does minification, and a bunch of other things. However when I benchmarked it didn't make enough of a difference to bother with, given my site was already well optimized. For sites that aren't optimized already it would probably make a significant difference. There's a tutorial on how to get Nginx/Pagespeed working here. You have ...


1

Reading from the restored volume should be sufficient. When you create a volume from an existing snapshot, it loads lazily in the background so that you can begin using them right away. If you access a piece of data that hasn't been loaded yet, the volume immediately downloads the requested data from Amazon S3, and then continues loading the rest of the ...


0

The Windows Performance Toolkit which is included in the ADK can log many performance metrics over an extended period of time, including per-file I/O statistics. http://go.microsoft.com/fwlink/p/?LinkId=526740


-1

Enabling Intel VT makes CPU hotter, I've had a Desktop and Laptop that have had this behaviour, both with stock CPU coolers. I'm refering to Home Computers but it's the same feature. I know that AMD-V comes enabled by default, but I don't know if it makes CPU hotter.


4

First of all - backups. RAID is not a replacement. Second - I would not go with hardware raid any more. On Linux you have ZFS, on Windows you have storage spaces. Both can provide a Raid 6 level security, but both can handle rebuilds a lot more gracefully. And both are cheaper - the cost of a decent hardware raid controller is not trivial. It was worth it ...


1

I guess you're out of luck. Why can't I get a larger heap with the 32-bit JVM? The maximum theoretical heap limit for the 32-bit JVM is 4G. Due to various additional constraints such as available swap, kernel address space usage, memory fragmentation, and VM overhead, in practice the limit can be much lower. On most modern 32-bit Windows ...


1

So, I've got a cause, but it's not really a root cause: mpstat relies on /proc/stat for its cpu metrics. pidstat relies on /proc/<pid>/stat. These files at times disagree about the number of kernel jiffies observed in a particular length of time, which I suppose makes sense as we're talking about such a increments of time resolution: 17:06:41 ...


0

The main difference between your two choices is 3 year warranty vs 5 years and SOHO usage area vs. certification for datacenters. In terms of raw I/O performance, there is not much of a difference between 5400/5900 and 7200 rpm S-ATA disks nowadays (that was different a few years ago). In the end therefore the array with more disks will very likely show a ...


9

This is hard to answer directly, because the answer is very much 'it depends on your caching'. You see - hard disks are slow. Very slow. But actually, most of the time you don't notice, if you have a decent cache, and a reasonably predicable access pattern. Read caching means recently accessed things will be coming off cache. Write caching means that ...


4

Based on calculations from RAID Performance Calculator, eight 5400rpm disks will yield about 304 IO/s. Whereas four 7200rpm disks will yield about 210 IO/s.



Top 50 recent answers are included