Search Results

Search found 5786 results on 232 pages for 'fast'.

facebook twitter digg google delicious technorati stumbleupon myspace wordpress linkedin gmail igoogle windows live tumblr viadeo yahoo buzz yahoo mail yahoo bookmarks favorites email print

Page 23/232 | < Previous Page | 19 20 21 22 23 24 25 26 27 28 29 30  | Next Page >

  • How fast can a Windows 2008 CIFS client write to SAMBA server using 10Gb NIC?

    - by one_bsd_guy
    We are experiencing a performance problem using Windows 2008 CIFS client. We have a FreeNAS server that delivers 1.3GB/s on ZFS write. We have 10Gb network connecting NAS server and CIFS clients. Using two Linux CIFS clients, we can get around 1.2GB/s. But windows 2008 clients can only give us 400MB/s. Is that the best a Windows 2008 client can deliver or we do have a poorly configured Windows client? Much appreciated.

    Read the article

  • How does Google manage to serve results so fast?

    - by Quintin Par
    I am building an autocomplete functionality for my site and the Google instant results are my benchmark. When I look at Google, the 50-60 ms response time baffle me. They look insane. In comparison here’s how mine looks like. To give you an idea my results are cached on the load balancer and served from a machine that has httpd slowstart and initcwnd fixed. My site is also behind cloudflare From a server side perspective I don’t think I can do anything more. Can someone help me take this 500 ms response time to 60ms? What more should I be doing to achieve Google level performance?

    Read the article

  • Is there a way to manage browser [chrome] session cookies with a fast toggle for multiple users?

    - by ResoluteHiker
    I feel I should go into more detail. My wife and I share a laptop and browse email. Currently we keep coming back to each others gmail accounts, having to log out and log back in. Are there any good extensions or addons that would allow us to toggle back and forth between these? This does not necessarily need to apply to just gmail but include any cookie, session, etc. I'd be willing to use Firefox if such an extension exists on it as well. Much appreciated!

    Read the article

  • What is the most platform- and Python-version-independent way to make a fast loop for use in Python?

    - by Statto
    I'm writing a scientific application in Python with a very processor-intensive loop at its core. I would like to optimise this as far as possible, at minimum inconvenience to end users, who will probably use it as an uncompiled collection of Python scripts, and will be using Windows, Mac, and (mainly Ubuntu) Linux. It is currently written in Python with a dash of NumPy, and I've included the code below. Is there a solution which would be reasonably fast which would not require compilation? This would seem to be the easiest way to maintain platform-independence. If using something like Pyrex, which does require compilation, is there an easy way to bundle many modules and have Python choose between them depending on detected OS and Python version? Is there an easy way to build the collection of modules without needing access to every system with every version of Python? Does one method lend itself particularly to multi-processor optimisation? (If you're interested, the loop is to calculate the magnetic field at a given point inside a crystal by adding together the contributions of a large number of nearby magnetic ions, treated as tiny bar magnets. Basically, a massive sum of these.) # calculate_dipole # ------------------------- # calculate_dipole works out the dipole field at a given point within the crystal unit cell # --- # INPUT # mu = position at which to calculate the dipole field # r_i = array of atomic positions # mom_i = corresponding array of magnetic moments # --- # OUTPUT # B = the B-field at this point def calculate_dipole(mu, r_i, mom_i): relative = mu - r_i r_unit = unit_vectors(relative) #4pi / mu0 (at the front of the dipole eqn) A = 1e-7 #initalise dipole field B = zeros(3,float) for i in range(len(relative)): #work out the dipole field and add it to the estimate so far B += A*(3*dot(mom_i[i],r_unit[i])*r_unit[i] - mom_i[i]) / sqrt(dot(relative[i],relative[i]))**3 return B

    Read the article

  • Will Learning C++ Help for Building Fast/No-Additional-Requirements Desktop Applications?

    - by vito
    Will learning C++ help me build native applications with good speed? Will it help me as a programmer, and what are the other benefits? The reason why I want to learn C++ is because I'm disappointed with the UI performances of applications built on top of JVM and .NET. They feel slow, and start slow too. Of course, a really bad programmer can create a slower and sluggish application using C++ too, but I'm not considering that case. One of my favorite Windows utility application is Launchy. And in the Readme.pdf file, the author of the program wrote this: 0.6 This is the first C++ release. As I became frustrated with C#’s large .NET framework requirements and users lack of desire to install it, I decided to switch back to the faster language. I totally agree with the author of Launchy about the .NET framework requirement or even a JRE requirement for desktop applications. Let alone the specific version of them. And some of the best and my favorite desktop applications don't need .NET or Java to run. They just run after installing. Are they mostly built using C++? Is C++ the only option for good and fast GUI based applications? And, I'm also very interested in hearing the other benefits of learning C++.

    Read the article

  • Fast, Unicode-capable, cross-platform programmer's text editor that shows invisibles like ZWSP?

    - by Roger_S
    Our publishing workflow includes Windows and Linux machines (there are some Macs too, but not in the critical-path workflow). Many texts include both English and Khmer and are marked-up in XML. XML Copy Editor is the best cross-platform open-source XML editor I've discovered. It utilizes the Scintilla editing component, which is generally good with Unicode but which does not enable non-printing or invisible characters like U+200B (zero-width space) and U+200C (zero-width non-joiner) to be displayed. Khmer does not separate words with a space character as Western languages do, so ZWSP is used in electronic texts to enable applications to break lines easily. Ideally I'd edit the markup and the content in a single editor, but XML awareness is less important at times than being able to display invisibles. (OpenOffice.org Writer and Microsoft Word are the only two apps I know that will display ZWSP. They are not suitable for the markup and text manipulations that need to be done to prepare manuscripts for publication, unfortunately, although I guess they're fine for authoring.) I tried out a promising editor last week, but a search-and-replace regex operation that took under a second in TextPad 4.7.3 lasted over twenty seconds. So I want to mention that speed and the ability to handle large (up to 150mb) files is also a concern. Is there a good, fast, free or not too expensive text editor, with versions on Windows and Linux and maybe mac too, Unicode-aware and capable of displaying invisibles like ZWSP? That has syntax highlighting, can handle large files and is customizable enough that I won't tear my hair out in frustration? Thanks, Roger_S

    Read the article

  • Why does Perl's Devel::LeakTrace::Fast point to blank files and evals?

    - by kt
    I am using Devel::LeakTrace::Fast to debug a memory leak in a perl script designed as a daemon which runs an infinite loop with sleeps until interrupted. I am having some trouble both reading the output and finding documentation to help me understand the output. The perldoc doesn't contain much information on the output. Most of it makes sense, such as pointing to globals in DBI. Intermingled with the output, however, are several leaked SV(<LOCATION>) from (eval #) line # Where the numbers are numbers and <LOCATION> is a location in memory. The script itself is not using eval at any point - I have not investigated each used module to see if evals are present. Mostly what I want to know is how to find these evals (if possible). I also find the following entries repeated over and over again leaked SV(<LOCATION>) from line # Where line # is always the same #. Not very helpful in tracking down what file that line is in.

    Read the article

  • Fast and efficient way to read a space separated file of numbers into an array?

    - by John_Sheares
    I need a fast and efficient method to read a space separated file with numbers into an array. The files are formatted this way: 4 6 1 2 3 4 5 6 2 5 4 3 21111 101 3 5 6234 1 2 3 4 2 33434 4 5 6 The first row is the dimension of the array [rows columns]. The lines following contain the array data. The data may also be formatted without any newlines like this: 4 6 1 2 3 4 5 6 2 5 4 3 21111 101 3 5 6234 1 2 3 4 2 33434 4 5 6 I can read the first line and initialize an array with the row and column values. Then I need to fill the array with the data values. My first idea was to read the file line by line and use the split function. But the second format listed gives me pause, because the entire array data would be loaded into memory all at once. Some of these files are in the 100 of MBs. The second method would be to read the file in chunks and then parse them piece by piece. Maybe somebody else has a better a way of doing this?

    Read the article

  • How can I combine several Expressions into a fast method?

    - by chillitom
    Suppose I have the following expressions: Expression<Action<T, StringBuilder>> expr1 = (t, sb) => sb.Append(t.Name); Expression<Action<T, StringBuilder>> expr2 = (t, sb) => sb.Append(", "); Expression<Action<T, StringBuilder>> expr3 = (t, sb) => sb.Append(t.Description); I'd like to be able to compile these into a method/delegate equivalent to the following: void Method(T t, StringBuilder sb) { sb.Append(t.Name); sb.Append(", "); sb.Append(t.Description); } What is the best way to approach this? I'd like it to perform well, ideally with performance equivalent to the above method. UPDATE So, whilst it appears that there is no way to do this directly in C#3 is there a way to convert an expression to IL so that I can use it with System.Reflection.Emit?

    Read the article

  • What is the meaning of the sentence "we wanted it to be compiled so it’s not burning CPU doing the wrong stuff."

    - by user2434
    I was reading this article. It has the following paragraph. And did Scala turn out to be fast? Well, what’s your definition of fast? About as fast as Java. It doesn’t have to be as fast as C or Assembly. Python is not significantly faster than Ruby. We wanted to do more with fewer machines, taking better advantage of concurrency; we wanted it to be compiled so it’s not burning CPU doing the wrong stuff. I am looking for the meaning of the last sentence. How will interpreted language make the CPU do "wrong" stuff ?

    Read the article

  • Unable to enable wireless on a Vostro 2520

    - by Joe
    I have a Vostro 2520 and not sure how to enable wireless on my machine. The details are given below, would appreciate any pointers to resolving this issue. lsmod returns Module Size Used by ath9k 132390 0 ath9k_common 14053 1 ath9k ath9k_hw 411151 2 ath9k,ath9k_common ath 24067 3 ath9k,ath9k_common,ath9k_hw b43 365785 0 mac80211 506816 2 ath9k,b43 cfg80211 205544 4 ath9k,ath,b43,mac80211 bcma 26696 1 b43 ssb 52752 1 b43 ndiswrapper 282628 0 ums_realtek 18248 0 usb_storage 49198 1 ums_realtek uas 18180 0 snd_hda_codec_hdmi 32474 1 snd_hda_codec_cirrus 24002 1 joydev 17693 0 parport_pc 32866 0 ppdev 17113 0 rfcomm 47604 0 bnep 18281 2 bluetooth 180104 10 rfcomm,bnep psmouse 97362 0 dell_wmi 12681 0 sparse_keymap 13890 1 dell_wmi snd_hda_intel 33773 3 snd_hda_codec 127706 3 snd_hda_codec_hdmi,snd_hda_codec_cirrus,snd_hda_intel snd_hwdep 13668 1 snd_hda_codec snd_pcm 97188 3 snd_hda_codec_hdmi,snd_hda_intel,snd_hda_codec snd_seq_midi 13324 0 snd_rawmidi 30748 1 snd_seq_midi snd_seq_midi_event 14899 1 snd_seq_midi snd_seq 61896 2 snd_seq_midi,snd_seq_midi_event snd_timer 29990 2 snd_pcm,snd_seq snd_seq_device 14540 3 snd_seq_midi,snd_rawmidi,snd_seq wmi 19256 1 dell_wmi snd 78855 16 snd_hda_codec_hdmi,snd_hda_codec_cirrus,snd_hda_intel,snd_hda_codec,snd_hwdep,snd_pcm,snd_rawmidi,snd_seq,snd_timer,snd_seq_device mac_hid 13253 0 i915 473240 3 drm_kms_helper 46978 1 i915 uvcvideo 72627 0 drm 242038 4 i915,drm_kms_helper videodev 98259 1 uvcvideo soundcore 15091 1 snd dell_laptop 18119 0 dcdbas 14490 1 dell_laptop i2c_algo_bit 13423 1 i915 v4l2_compat_ioctl32 17128 1 videodev snd_page_alloc 18529 2 snd_hda_intel,snd_pcm video 19596 1 i915 serio_raw 13211 0 mei 41616 0 lp 17799 0 parport 46562 3 parport_pc,ppdev,lp r8169 62099 0 sudo lshw -class network *-network UNCLAIMED description: Network controller product: Broadcom Corporation vendor: Broadcom Corporation physical id: 0 bus info: pci@0000:07:00.0 version: 01 width: 64 bits clock: 33MHz capabilities: pm msi pciexpress bus_master cap_list configuration: latency=0 resources: memory:f7c00000-f7c07fff *-network description: Ethernet interface product: RTL8111/8168B PCI Express Gigabit Ethernet controller vendor: Realtek Semiconductor Co., Ltd. physical id: 0 bus info: pci@0000:09:00.0 logical name: eth0 version: 07 serial: 78:45:c4:a3:aa:65 size: 100Mbit/s capacity: 1Gbit/s width: 64 bits clock: 33MHz capabilities: pm msi pciexpress msix vpd bus_master cap_list ethernet physical tp mii 10bt 10bt-fd 100bt 100bt-fd 1000bt 1000bt-fd autonegotiation configuration: autonegotiation=on broadcast=yes driver=r8169 driverversion=2.3LK-NAPI duplex=full firmware=rtl8168e-3_0.0.4 03/27/12 ip=192.168.1.5 latency=0 link=yes multicast=yes port=MII speed=100Mbit/s resources: irq:41 ioport:e000(size=256) memory:f0004000-f0004fff memory:f0000000-f0003fff rfkill list all 0: dell-wifi: Wireless LAN Soft blocked: yes Hard blocked: yes 1: dell-bluetooth: Bluetooth Soft blocked: yes Hard blocked: yes Output of lspci > 00:00.0 Host bridge: Intel Corporation Ivy Bridge DRAM Controller (rev > 09) 00:02.0 VGA compatible controller: Intel Corporation Ivy Bridge > Graphics Controller (rev 09) 00:16.0 Communication controller: Intel > Corporation Panther Point MEI Controller #1 (rev 04) 00:1a.0 USB > controller: Intel Corporation Panther Point USB Enhanced Host > Controller #2 (rev 04) 00:1b.0 Audio device: Intel Corporation Panther > Point High Definition Audio Controller (rev 04) 00:1c.0 PCI bridge: > Intel Corporation Panther Point PCI Express Root Port 1 (rev c4) > 00:1c.3 PCI bridge: Intel Corporation Panther Point PCI Express Root > Port 4 (rev c4) 00:1c.5 PCI bridge: Intel Corporation Panther Point > PCI Express Root Port 6 (rev c4) 00:1d.0 USB controller: Intel > Corporation Panther Point USB Enhanced Host Controller #1 (rev 04) > 00:1f.0 ISA bridge: Intel Corporation Panther Point LPC Controller > (rev 04) 00:1f.2 SATA controller: Intel Corporation Panther Point 6 > port SATA Controller [AHCI mode] (rev 04) 00:1f.3 SMBus: Intel > Corporation Panther Point SMBus Controller (rev 04) 07:00.0 Network > controller: Broadcom Corporation Device 4365 (rev 01) 09:00.0 Ethernet > controller: Realtek Semiconductor Co., Ltd. RTL8111/8168B PCI Express > Gigabit Ethernet controller (rev 07) Output of lspci -v 0:00.0 Host bridge: Intel Corporation Ivy Bridge DRAM Controller (rev 09) Subsystem: Dell Device 0558 Flags: bus master, fast devsel, latency 0 Capabilities: <access denied> Kernel driver in use: agpgart-intel 00:02.0 VGA compatible controller: Intel Corporation Ivy Bridge Graphics Controller (rev 09) (prog-if 00 [VGA controller]) Subsystem: Dell Device 0558 Flags: bus master, fast devsel, latency 0, IRQ 43 Memory at f7800000 (64-bit, non-prefetchable) [size=4M] Memory at e0000000 (64-bit, prefetchable) [size=256M] I/O ports at f000 [size=64] Expansion ROM at <unassigned> [disabled] Capabilities: <access denied> Kernel driver in use: i915 Kernel modules: i915 00:16.0 Communication controller: Intel Corporation Panther Point MEI Controller #1 (rev 04) Subsystem: Dell Device 0558 Flags: bus master, fast devsel, latency 0, IRQ 42 Memory at f7d0a000 (64-bit, non-prefetchable) [size=16] Capabilities: <access denied> Kernel driver in use: mei Kernel modules: mei 00:1a.0 USB controller: Intel Corporation Panther Point USB Enhanced Host Controller #2 (rev 04) (prog-if 20 [EHCI]) Subsystem: Dell Device 0558 Flags: bus master, medium devsel, latency 0, IRQ 16 Memory at f7d08000 (32-bit, non-prefetchable) [size=1K] Capabilities: <access denied> Kernel driver in use: ehci_hcd 00:1b.0 Audio device: Intel Corporation Panther Point High Definition Audio Controller (rev 04) Subsystem: Dell Device 0558 Flags: bus master, fast devsel, latency 0, IRQ 44 Memory at f7d00000 (64-bit, non-prefetchable) [size=16K] Capabilities: <access denied> Kernel driver in use: snd_hda_intel Kernel modules: snd-hda-intel 00:1c.0 PCI bridge: Intel Corporation Panther Point PCI Express Root Port 1 (rev c4) (prog-if 00 [Normal decode]) Flags: bus master, fast devsel, latency 0 Bus: primary=00, secondary=04, subordinate=04, sec-latency=0 Capabilities: <access denied> Kernel driver in use: pcieport Kernel modules: shpchp 00:1c.3 PCI bridge: Intel Corporation Panther Point PCI Express Root Port 4 (rev c4) (prog-if 00 [Normal decode]) Flags: bus master, fast devsel, latency 0 Bus: primary=00, secondary=07, subordinate=07, sec-latency=0 Memory behind bridge: f7c00000-f7cfffff Capabilities: <access denied> Kernel driver in use: pcieport Kernel modules: shpchp 00:1c.5 PCI bridge: Intel Corporation Panther Point PCI Express Root Port 6 (rev c4) (prog-if 00 [Normal decode]) Flags: bus master, fast devsel, latency 0 Bus: primary=00, secondary=09, subordinate=09, sec-latency=0 I/O behind bridge: 0000e000-0000efff Prefetchable memory behind bridge: 00000000f0000000-00000000f00fffff Capabilities: <access denied> Kernel driver in use: pcieport Kernel modules: shpchp 00:1d.0 USB controller: Intel Corporation Panther Point USB Enhanced Host Controller #1 (rev 04) (prog-if 20 [EHCI]) Subsystem: Dell Device 0558 Flags: bus master, medium devsel, latency 0, IRQ 23 Memory at f7d07000 (32-bit, non-prefetchable) [size=1K] Capabilities: <access denied> Kernel driver in use: ehci_hcd 00:1f.0 ISA bridge: Intel Corporation Panther Point LPC Controller (rev 04) Subsystem: Dell Device 0558 Flags: bus master, medium devsel, latency 0 Capabilities: <access denied> Kernel modules: iTCO_wdt 00:1f.2 SATA controller: Intel Corporation Panther Point 6 port SATA Controller [AHCI mode] (rev 04) (prog-if 01 [AHCI 1.0]) Subsystem: Dell Device 0558 Flags: bus master, 66MHz, medium devsel, latency 0, IRQ 40 I/O ports at f0b0 [size=8] I/O ports at f0a0 [size=4] I/O ports at f090 [size=8] I/O ports at f080 [size=4] I/O ports at f060 [size=32] Memory at f7d06000 (32-bit, non-prefetchable) [size=2K] Capabilities: <access denied> Kernel driver in use: ahci 00:1f.3 SMBus: Intel Corporation Panther Point SMBus Controller (rev 04) Subsystem: Dell Device 0558 Flags: medium devsel, IRQ 11 Memory at f7d05000 (64-bit, non-prefetchable) [size=256] I/O ports at f040 [size=32] Kernel modules: i2c-i801 07:00.0 Network controller: Broadcom Corporation Device 4365 (rev 01) Subsystem: Dell Device 0016 Flags: bus master, fast devsel, latency 0, IRQ 10 Memory at f7c00000 (64-bit, non-prefetchable) [size=32K] Capabilities: <access denied> 09:00.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL8111/8168B PCI Express Gigabit Ethernet controller (rev 07) Subsystem: Dell Device 0558 Flags: bus master, fast devsel, latency 0, IRQ 41 I/O ports at e000 [size=256] Memory at f0004000 (64-bit, prefetchable) [size=4K] Memory at f0000000 (64-bit, prefetchable) [size=16K] Capabilities: <access denied> Kernel driver in use: r8169 Kernel modules: r8169

    Read the article

  • How fast is Berkeley DB SQL compared to SQLite?

    - by dan04
    Oracle recently released a Berkeley DB back-end to SQLite. I happen to have a hundreds-of-megabytes SQLite database that could very well benefit from "improved performance, concurrency, scalability, and reliability", but Oracle's site appears to lack any measurements of the improvements. Has anyone here done some benchmarking?

    Read the article

  • Is there a fast way to jump to element using XMLReader?

    - by Derk
    I am using XMLReader to read a large XML file with about 1 million elements on the level I am reading from. However, I've calculated it will take over 10 seconds when I jump to -for instance- element 500.000 using XMLReader::next ([ string $localname ] ) or XMLReader::read ( void ) This is not very usable. Is there a faster way to do this?

    Read the article

  • Fast JSON serialization (and comparison with Pickle) for cluster computing in Python?

    - by user248237
    I have a set of data points, each described by a dictionary. The processing of each data point is independent and I submit each one as a separate job to a cluster. Each data point has a unique name, and my cluster submission wrapper simply calls a script that takes a data point's name and a file describing all the data points. That script then accesses the data point from the file and performs the computation. Since each job has to load the set of all points only to retrieve the point to be run, I wanted to optimize this step by serializing the file describing the set of points into an easily retrievable format. I tried using JSONpickle, using the following method, to serialize a dictionary describing all the data points to file: def json_serialize(obj, filename, use_jsonpickle=True): f = open(filename, 'w') if use_jsonpickle: import jsonpickle json_obj = jsonpickle.encode(obj) f.write(json_obj) else: simplejson.dump(obj, f, indent=1) f.close() The dictionary contains very simple objects (lists, strings, floats, etc.) and has a total of 54,000 keys. The json file is ~20 Megabytes in size. It takes ~20 seconds to load this file into memory, which seems very slow to me. I switched to using pickle with the same exact object, and found that it generates a file that's about 7.8 megabytes in size, and can be loaded in ~1-2 seconds. This is a significant improvement, but it still seems like loading of a small object (less than 100,000 entries) should be faster. Aside from that, pickle is not human readable, which was the big advantage of JSON for me. Is there a way to use JSON to get similar or better speed ups? If not, do you have other ideas on structuring this? (Is the right solution to simply "slice" the file describing each event into a separate file and pass that on to the script that runs a data point in a cluster job? It seems like that could lead to a proliferation of files). thanks.

    Read the article

  • How can I combine sequential expression trees into a fast method?

    - by chillitom
    Suppose I have the following expressions: Expression<Action<T, StringBuilder>> expr1 = (t, sb) => sb.Append(t.Name); Expression<Action<T, StringBuilder>> expr2 = (t, sb) => sb.Append(", "); Expression<Action<T, StringBuilder>> expr3 = (t, sb) => sb.Append(t.Description); I'd like to be able to compile these into a method/delegate equivalent to the following: void Method(T t, StringBuilder sb) { sb.Append(t.Name); sb.Append(", "); sb.Append(t.Description); } What is the best way to approach this? I'd like it to perform well, ideally with performance equivalent to the above method.

    Read the article

  • Blog Engine .NET with XML Data Storage How is it so fast ?

    - by Kubi
    I am looking for a way to embed a blog engine into my own application and I am too curious about Blog Engine algorithm. This may not be the correct place to ask but, How is that possible to store blog entry data in an XML file like BlogEngine.Net with Default Configuration. It must be getting slower everyday while the file is getting larger and larger. I am wondering the algorithm behind that. Is it loading with a different way ? Or Am I wrong with the time estimation ? I know it is open source but I thought it would be better to see a discussion here for some others might be thinking the same and this thread can be a reference.

    Read the article

  • Fast sign in C++ float...are there any platform dependencies in this code?

    - by Patrick Niedzielski
    Searching online, I have found the following routine for calculating the sign of a float in IEEE format. This could easily be extended to a double, too. // returns 1.0f for positive floats, -1.0f for negative floats, 0.0f for zero inline float fast_sign(float f) { if (((int&)f & 0x7FFFFFFF)==0) return 0.f; // test exponent & mantissa bits: is input zero? else { float r = 1.0f; (int&)r |= ((int&)f & 0x80000000); // mask sign bit in f, set it in r if necessary return r; } } (Source: ``Fast sign for 32 bit floats'', Peter Schoffhauzer) I am weary to use this routine, though, because of the bit binary operations. I need my code to work on machines with different byte orders, but I am not sure how much of this the IEEE standard specifies, as I couldn't find the most recent version, published this year. Can someone tell me if this will work, regardless of the byte order of the machine? Thanks, Patrick

    Read the article

  • Infiniband: a highperformance network fabric - Part I

    - by Karoly Vegh
    Introduction:At the OpenWorld this year I managed to chat with interesting people again - one of them answering Infiniband deepdive questions with ease by coffee turned out to be one of Oracle's IB engineers, Ted Kim, who actually actively participates in the Infiniband Trade Association and integrates Oracle solutions with this highspeed network. This is why I love attending OOW. He granted me an hour of his time to talk about IB. This post is mostly based on that tech interview.Start of the actual post: Traditionally datatransfer between servers and storage elements happens in networks with up to 10 gigabit/seconds or in SANs with up to 8 gbps fiberchannel connections. Happens. Well, data rather trickles through.But nowadays data amounts grow well over the TeraByte order of magnitude, and multisocket/multicore/multithread Servers hunger data that these transfer technologies just can't deliver fast enough, causing all CPUs of this world do one thing at the same speed - waiting for data. And once again, I/O is the bottleneck in computing. FC and Ethernet can't keep up. We have half-TB SSDs, dozens of TB RAM to store data to be modified in, but can't transfer it. Can't backup fast enough, can't replicate fast enough, can't synchronize fast enough, can't load fast enough. The bad news is, everyone is used to this, like back in the '80s everyone was used to start compile jobs and go for a coffee. Or on vacation. The good news is, there's an alternative. Not so-called "bleeding-edge" 8gbps, but (as of now) 56. Not layers of overhead, but low latency. And it is available now. It has been for a while, actually. Welcome to the world of Infiniband. Short history:Infiniband was born as a result of joint efforts of HPAQ, IBM, Intel, Sun and Microsoft. They planned to implement a next-generation I/O fabric, in the 90s. In the 2000s Infiniband (from now on: IB) was quite popular in the high-performance computing field, powering most of the top500 supercomputers. Then in the middle of the decade, Oracle realized its potential and used it as an interconnect backbone for the first Database Machine, the first Exadata. Since then, IB has been booming, Oracle utilizes and supports it in a large set of its HW products, it is the backbone of the famous Engineered Systems: Exadata, SPARC SuperCluster, Exalogic, OVCA and even the new DB backup/recovery box. You can also use it to make servers talk highspeed IP to eachother, or to a ZFS Storage Appliance. Following Oracle's lead, even IBM has jumped the wagon, and leverages IB in its PureFlex systems, their first InfiniBand Machines.IB Structural Overview: If you want to use IB in your servers, the first thing you will need is PCI cards, in IB terms Host Channel Adapters, or HCAs. Just like NICs for Ethernet, or HBAs for FC. In these you plug an IB cable, going to an IB switch providing connection to other IB HCAs. Of course you're going to need drivers for those in your OS. Yes, these are long-available for Solaris and Linux. Now, what protocols can you talk over IB? There's a range of choices. See, IB isn't accepting package loss like Ethernet does, and hence doesn't need to rely on TCP/IP as a workaround for resends. That is, you still can run IP over IB (IPoIB), and that is used in various cases for control functionality, but the datatransfer can run over more efficient protocols - like native IB. About PCI connectivity: IB cards, as you see are fast. They bring low latency, which is just as important as their bandwidth. Current IB cards run at 56 gbit/s. That is slightly more than double of the capacity of a PCI Gen2 slot (of ~25 gbit/s). And IB cards are equipped usually with two ports - that is, altogether you'd need 112 gbit/s PCI slots, to be able to utilize FDR IB cards in an active-active fashion. PCI Gen3 slots provide you with around ~50gbps. This is why the most IB cards are configured in an active-standby way if both ports are used. Once again the PCI slot is the bottleneck. Anyway, the new Oracle servers are equipped with Gen3 PCI slots, an the new IB HCAs support those too. Oracle utilizes the QDR HCAs, running at 40gbp/s brutto, which translates to a 32gbp/s net traffic due to the 10:8 signal-to-data information ratio. Consolidation techniques: Technology never stops to evolve. Mellanox is working on the 100 gbps (EDR) version already, which will be optical, since signal technology doesn't allow EDR to be copper. Also, I hear you say "100gbps? I will never use/need that much". Are you sure? Have you considered consolidation scenarios, where (for example with Oracle Virtual Network) you could consolidate your platform to a high densitiy virtualized solution providing many virtual 10gbps interfaces through that 100gbps? Technology never stops to evolve. I still remember when a 10mbps network was impressively fast. Back in those days, 16MB of RAM was a lot. Now we usually run servers with around 100.000 times more RAM. If network infrastrucure speends could grow as fast as main memory capacities, we'd have a different landscape now :) You can utilize SRIOV as well for consolidation. That is, if you run LDoms (aka Oracle VM Server for SPARC) you do not have to add physical IB cards to all your guest LDoms, and you do not need to run VIO devices through the hypervisor either (avoiding overhead). You can enable SRIOV on those IB cards, which practically virtualizes the PCI bus, and you can dedicate Physical- and Virtual Functions of the virtualized HCAs as native, physical HW devices to your guests. See Raghuram's excellent post explaining SRIOV. SRIOV for IB is supported since LDoms 3.1.  This post is getting lengthier, so I will rename it to Part I, and continue it in a second post. 

    Read the article

  • What's the easiest/fast way to get my website up and running on the web?

    - by ggfan
    This is probably a really really beginner's question, but I would like to know what's the fastest way to get my site on the web so that people can start using it. I'm learning everything about programming out of books and at home so I don't have much experience. --Before I go to like godaddy.com or such site to get a domain name, is there any free sites that would allow me to upload my site so users can use it? I have html,css,php,mysql,javascipt in my scripts so I don't think many sites allow free uploads with such languages. --If I can't find a free site, is there any good places to get a domain name and web hosting that supports most languages at a low price? (doesn't have to be professional hosting because I am still a beginner) --If I go to say godaddy.com and get their webhosting and domain name, would I be allowed to run php,mysql,python,java on it? (I looked at some hosting sites and most only allow php/mysql)

    Read the article

facebook twitter digg google delicious technorati stumbleupon myspace wordpress linkedin gmail igoogle windows live tumblr viadeo yahoo buzz yahoo mail yahoo bookmarks favorites email print

< Previous Page | 19 20 21 22 23 24 25 26 27 28 29 30  | Next Page >