From: ** http://www.micro * ** /ntserver/nts/news/msnw/LinuxMyths.asp * ** * PP ** Linux Myths * P PP ** Posted: Octobe * P PP ** * PP P ** With all the * P PP P AA ** inux as an operating system, it's important t * P PP A AA ** look at the reality. First, it's worth noting tha * P A AA ** system. Linux fundamentally relies on ** AA A ** ology and architecture. Linux was not des ** A AA A N ** symmetrical multiprocessi ** A AA NN ** (GUI), asynchronous I/O, fine-grained se ** A A N N N ** ant characteristics of a modern operating ** A N N N ** ions mean that as customers look for ** A N N N TT ** oy scalable, secure, and robust applicatio ** N N N TT ** the hype. ** N N N T TT ** Myth: Linux performs bett ** NN T ** ** N T SS ** Reality: Windows NT 4.0 Outpe ** T S SS ** Workloads ** T S SS * ** T S * The Linux community claims to have ** SS * scalability in the latest versions of ** S * , however it's clea that Linux remains inferior to the Wind ** SS S * g system. ** SS S * For File and Print services, accordi ** SS * ent tests conducted by PC Week Labs, the Windows ** * rating system delivers 52 percent better perfor ** * a single processor system and 110 percent better pe ** ance on a 4-way system than similarly configured singl processor and 4-way Linux/S BA systems. For Web serv ** the same PC Week tests showed Windows NT 4.0 with In * ** nformation Server 4.0 delivers 41 percent better per * ** a single processor system and 125 percent b * PP ** nce on a 4-way system than Linux and Apache. * P PP ** For e-c * P PP ** ing secure sockets (SSL), recent PC Mag * PP P ** ows NT 4.0 with Internet Infor * P PP P AA ** approximately five times the perf * P PP A AA ** Stronghold. For * P A AA ** Business applications, Win ** AA A ** of 40,368 tpmC at a cost ** A AA A N ** paq 8-Way Pentium III XEON p ** A AA NN ** y leading price/pe ** A A N N N ** on Processing Performanc ** A N N N ** indows NT can deliver ** A N N N TT ** uty transaction processing. It ** N N N TT ** s not a single TPC result on an ** N N N T TT ** erefore Linux has yet to d ** NN T ** database server. ** N T SS ** Linux performance and ** T S SS ** ited in the 2.2 Kernel. Linux on ** T S SS * RAM on the x86 architecture,1 ** T S * NT 4.0. The largest file size L ** SS * us 16 terabytes (TB) for Windows NT ** S * e is limited to 128 MB. In addition, ** SS S * rt many of the modern operating system fea ** SS S * s NT 4.0 has pioneered such as asynchronous I ** SS * ports, and fine-grained kernel locks. These a ** * constraints limit the ability of Linux to scale well ** * o processors. The Linux community continues to promise m ** SMP and performance improvements. They have been pro sing these since the dev opment of the 2.0 Kernel in 1996. Delivering a scalable sys ** s a complex task and it's not clear that the Linux commu * ** solve these issues easily or quickly. As D. H. Brow * ** s noted in a recent technical report,2 the Linux 2.2 * PP ** s in the early stages of providing a tuned SM * P PP ** * P PP ** Myth: Linux i * PP P ** ndows NT * P PP P AA ** Reality: Li * P PP A AA ** ints Rather than Anecdotal Stories * P A AA ** ** AA A ** The Linux c ** A AA A N ** as a stable and reliable operating sys ** A AA NN ** ta or metrics and very limited custome ** A A N N N ** ** A N N N ** Windows NT 4 ** A N N N TT ** tomer environments t ** N N N TT ** ustomers such as Barnes a ** N N N T TT ** go Stock Exchange, De ** NN T ** run mission-critical app ** N T SS ** Linux lacks a commerci ** T S SS ** This means that in the event ** T S SS * ower outage) data loss or corru ** T S * nt, the system must check the integr ** SS * ing system restart, a process that ** S * extended amount of time, especially on la ** SS S * require manual intervention to reconstruct ** SS S * There are no commercially proven clu ** SS * logies to provide High Availability for Linux. T ** * mmunity may point to numerous projects and small com ** * hat are aiming to deliver High Availability functi ** ty. D. H. Brown recently noted that these offerings remain i ature and largely unpro n in the demanding business world. There are no ** that provide uptime guarantees for Linux, unlike Wind * ** here Compaq, Data General, Hewlett-Pa * ** and Unisys provide 99.9 percent system-le * PP ** rantees for Windows NT-based servers. * P PP ** Myth: Linux is * P PP ** * PP P ** Reality: Fre * P PP P AA ** t Mean Low Total Cost of Ownership * P PP A AA ** * P A AA ** The Linux ** AA A ** or low-cost nature of Linux. It's import ** A AA A N ** st is only a small part of the overall decis ** A AA NN ** ** A A N N N ** The cost o ** A N N N ** all percentage of the overall ** A N N N TT ** eneral Windows NT has proven ** N N N TT ** han UNIX. Previous studies ** N N N T TT ** percent lower TCO than UNI ** NN T ** at Linux is significant ** N T SS ** UNIX when it comes to TCO. ** T S SS ** The very definition of L ** T S SS * means that commercial companies ** T S * by charging for services. There ** SS * ervices for Linux will be fee-based an ** S * at a premium. These costs have to be ** SS S * tal cost model. ** SS S * Linux is a UNIX-like operating syste ** SS * fore complex to configure and manage. Existing UNIX ** * find the transition to Linux easier but administr ** * r existing Windows®-based or Novell environments will ** it more difficult to handle the complexity of Linux. his retraining will add significa costs to Linux deployments. Linux is a h ** risk option than Windows NT. For example how many ce * ** engineers are there for Linux? How easy is it to find * ** velopment and support people for Linux? Who perfo * PP ** testing for Linux-based solutions? These fa * P PP ** ed to be taken into account when choosin * P PP ** r business. * PP P ** Myth: Linux * P PP P AA ** s NT * P PP A AA ** Reality: L * P A AA ** ** AA A ** All systems ** A AA A N ** however it's important to note that Linux us ** A AA NN ** riginal UNIX implementations ** A A N N N ** m the ground up to be secure. ** A N N N ** ** A N N N TT ** Linux only pro ** N N N TT ** directories. In contrast, every ** N N N T TT ** operating system data struct ** NN T ** d its use can be regulated as ** N T SS ** Linux security is all- ** T S SS ** delegate administrative privilege ** T S SS * rative capability must be made a ** T S * compromises best security pr ** SS * ows NT allows an administrator to del ** S * exceptionally fine-grained level ** SS S * Linux has not supported key securi ** SS S * standards. Every member of the Windows NT famil ** SS * s NT 3.5 has been evaluated at either a C2 ** * the U.S. Government's evaluation process or at a ** * alent level under the British Government's ITSEC proce ** n contrast, no Linux products are listed on the U.S. Govern nt's evaluated product list. Linux system ** nistrators must spend huge amounts of time understandi * ** atest Linux bugs and determining what to do about t * ** s made complex due to the fact that there isn * PP ** ocation for security issues to be reported and fixe * P PP ** crosoft provides a single security reposit * P PP ** and fixes of security related issues. Config * PP P ** quires an administrator to be an exper * P PP P AA ** e operating system and how comp * P PP A AA ** any part of the operating sys * P A AA ** erable to attack. Windows NT ** AA A ** inister with tools such as the ** A AA A N ** ** A AA NN ** Myth: Linux can ** A A N N N ** ** A N N N ** Reality: Linux Make ** A N N N TT ** ** N N N TT ** Linux as a desktop oper ** N N N T TT ** would end up with a system that has fe ** NN T ** o use and manage, and is less intuiti ** N T SS ** ** T S SS ** Linux does not provide s ** T S SS * hardware in use today; Win ** T S * s over 39,000 systems and devices o ** SS * ty List. Linux does not support i ** S * technologies such as Plug and Pl ** SS S * Management ** SS S * The complexity of the Linux operatin ** SS * cumbersome nature of the existing GUIs ** * retraining end-users a huge undertaking ** * d add significant cost ** Linux application support is very limited, m ning that customers end p having to build their own horizontal and vertical app ** ions. A recent report from Forrester Research highlighted * ** t that today 93 percent of enterprise ISVs develop ap * ** for Windows NT, while only 13 percent develop f * PP ** * P PP ** Summary * P PP ** The Linux ope * PP P ** itable for mainstream usage by business or * P PP P AA ** dows NT 4.0, customers can be confident i * P PP A AA ** t are scalable, secure, and reliable-- * P A AA ** manage. Linux clearly has a long way to go ** AA A ** 4.0. With the release of the Windows 200 ** A AA A N ** ds the technical superiority o ** A AA NN ** g that customers can deliver the next genera ** A A N N N ** siness challenges. ** A N N N ** More information ** A N N N TT ** ** N N N TT ** Customer Testimonials ** N N N T TT ** See how these leading com ** NN T ** yed Windows NT Server 4.0: ** N T SS ** ** T S SS ** Nasdaq ** T S SS * Barnes & Noble ** T S * Dell Computer Corp ** SS * The Boeing Company ** S * Chicago Stock Exchange ** SS S * ** SS S * Gartner Group Reports ** SS * New reports from Gartner raise important ques ** * the future role of Linux. ** * ** Will Linux Be Viable Competition for Windows esktops? "While we do ** view Linux as a serious competitor for Microsoft a * ** sktop, Linux will not disappear from the computing * ** hrough 2004." 1999 OS F * PP ** inux Face-Off "While L * P PP ** portant niche roles, it will not gain broad a * P PP ** itute for Unix and Windows in the enterp * PP P ** Red H * P PP P AA ** "We * P PP A AA ** or success in the Linux mar * P A AA ** s not a foregone conclusion, des ** AA A ** ** A AA A N ** Performance D ** A AA NN ** See Industry Be ** A A N N N ** Outperforms Linux ** A N N N ** ** A N N N TT ** Footnotes ** N N N TT ** 1. Siemens & SuSE annou ** N N N T TT ** xtend to 4 GB, although this is no ** NN T ** istributions. ** N T SS ** 2. Linux: How Good Is It? D. ** T S SS ** 9 ** T S SS * 3. Forrester Research, Software V ** T S * Aug. 31, 1999 ** SS * |--------------------------------------------------------------------| | This file has been borrocksed by Borrocks 95. | | (c) 1999 Scott A. Belmonte, Lachlan M. D. Cranswick | | http://www.ccp14.ac.uk/people/lachlan/borrocks/ | |--------------------------------------------------------------------|