Solaris System Resource Manager as It Is - Essay Example

Running Head: What sort Of entity is Solaris? What sort of entity is Solaris? of the of the What sort of entity is Solaris? In 1999, at the CMG Conference in Reno, Nevada, Dr. Gunther presented a paper entitled Solaris System Resource Manager: All I Ever Wanted Was My Unfair Advantage (And Why You Cant Get It) [1]. The traditional UNIX time-share (TS) schedulers have implemented a round-robin algorithm. While this algorithm ensures responsiveness to multiple interactive users, it favors users that run many jobs with short CPU usage.[2] The Solaris System Resource Manager (SRM) is a software tool that enables the allocation and control οf major system resources such as CPU, physical memory, virtual memory and number οf processes [3]. The system administrator can allocate CPU resources according to predefined shares as opposed to fixed percentages, which allows the system to dynamically apportion all available resources according to the relative proportion οf shares οf any current user. The SRM pro rates resource shares to users and groups and then adjusts CPU usage to meet the shares. Dr. Gunther presents two significant differences between the TS and SRM schedulers. The first is that the SRM guarantees a minimum percentage οf CPU, rather than a fixed percentage. The other difference is that when the allocations are changed dynamically, the SRM changes are not always immediately reflected the in percentages οf CPU time the users receive. Dr. Gunthers first example explains that if a user is awarded 10 οf 100 shares, that user receives a minimum οf 10% οf the CPU resources when the machine is busy. If the machine is only 50% active, the same user will receive double or 20% οf the CPU resources. This CPU usage is determined by an instantaneous and periodic sampling οf the usage to adjust the resource usage. Since the usage has to be sampled and adjusted, this causes a time lag between the allocation and the realization οf resources. In his second example, Gunther shows that if two users equally share the system resources, when one is not logged on, the SRM will adjust and allow the single user to use 100% οf the resources. When the second user logs on, the resources are split in half. For all οf the examples in this paper, the processes are assumed to be CPU-bound to present a least upper bound model οf resource usage. Learning Outcome The goal οf the SRM is to dynamically adjust each users CPU usage to reflect the ratio οf shares to which the user is entitled. Dr. Gunther uses a modeling tool called PDQ to demonstrate several capacity planning scenarios. The first scenario presents two small share users in one group. The data compares TS and SRM response times as well as comparisons οf SRM response times between scenarios. The user with fewer shares in the first scenario has a longer wait than with traditional TS schedulers. The wait is significantly longer for a small share user when a large share user is brought online in the second scenario. When two groups are active in the third scenario, the group with smaller shares suffers performance degradation. The fourth scenario presents results from all three groups being active. The groups with the largest number οf shares have significant performance improvement while the opposite occurs for the small share groups. Gunther points out that allowing a sudden swing in response times by an order οf magnitude or more is highly undesirable when allocating SRM shares and this has to be considered when a particular group is given too many resource shares. In addition, a single user from a different group can have a large impact on a separate group. Dr. Gunther recommends a method to use when setting the SRM tuning parameters. If on a particular system, the service demands and workload intensity are less relative to the case studies presented, Gunther predicts the SRM will performance will be better than predicted. If each user has more than one process executing at a time or if there is a great disparity in the work performed by different groups, the SRM performance will be worse than modeled. System administrators should measure their maximum CPU utilization under standard TS and note the response time. When setting the CPU shares under SRM, system administrators should allocate shares such that the entitlement ratio is at least equal to the maximum CPU utilization measured. In this paper, Solaris SRM is compared to the International Business Machines Corporation (IBM) older Multiple Virtual Storage (MVS) SRM and the newer MVS Work Load Manager (WLM). Gunther indicates that the Solaris SRM is close in functionality to the IBM SRM οf a decade ago. IBMs newer MVS WLM offers a new capability οf integrating CPU resources a user receives over time and making adjustments if the user is not receiving his fair share. The SRM implementation only looks at a snapshot οf CPU usage to make adjustments. The IBM WLM also looks at waiting time in the queue for a job and adjusts the position οf a job in the queue. In essence, the SRM and WLM have goals that are complements οf the CPU capacity spectrum where SRM uses CPU utilization and WLM uses delay or wait time. Gunther appears to view the SRM as a Solaris improvement over the TS, but Solaris still does not have a WLM-like implementation. Market Share One οf the biggest battles that occur between companies that operate for profit is for market share. Market share is the percentage οf industry sales οf a product or service that is controlled by a company. While a large market share is not always an indication οf profitability, it does provide a company with more opportunities to make a profit. In the information technology industry, the battle for operating system market share has been epic. Although it is common knowledge that Microsoft has dominated the market share since the late 1990’s, operating system market share cannot be fully understood without intricate analysis. Operating system market share can be summarizing by the following subjects: quantity shipped, server markets, and developer concentration. The most common aspect οf a company’s market share is the percentage οf total quantity shipped οf a particular product. This seems like a simple enough concept, but there is more to it than numbers. As οf August 2006, Windows platforms made up almost ninety percent οf the installed operating systems in use, with Windows XP chewing up over 75% οf the systems (Refuses Data, 2006). UNIX operating systems account for less than 0.5% οf installed operating systems and Linux accounts for 3.5%. These statistics would lead one to believe that Windows XP has 75% οf the market, but this cannot be assumed. The statistics οf installed operating systems are the result οf the culmination οf many years οf sales and are not an accurate representation οf the current market. These statistics do not depict variables such as computers that were sold with Windows XP and have since been changed to Linux or vice versa. Different companies calculate units shipped statistics in different ways. Microsoft uses the total number οf products shipped, whether or not they actually get used. This means that products that are sitting on store shelves unsold are included in Microsoft’s “units shipped” statistics (Petreley, 2003). Linux market share is based on surveys, commercial off the shelf boxes sold, and downloads. The two statistics are not comparable. Omitting or admitting systems such as cell phones and other handheld devices can also skew statistics. The server market is an important aspect οf market share due to the high cost οf server products. The server market is fast expanding, as web-servers become a must-have aspect οf business. Windows overall server market share was about 33% in 2005, while UNIX systems made up 36% οf the server market. Linux systems accounted for nine percent οf the server market (Woodie, 2005). Windows server revenue grew 15% in 2005, putting it at a pace that will overrun UNIX by 2007. Windows claimed 30% οf the web-server market as οf June 2006, up 4 % from May 2006. These statistics show that although Windows dominates the overall market share, Microsoft is still in heavy competition for big money markets such as the server market. IT developers are shifting their concentration from Windows development to Unix/Linux development. In a survey conducted by Evans Data Corporation, 50% οf developers surveyed focused on the development οf Windows products in 2003, while 40% οf the developers concentrated on Unix/Linux products (Petreley, 2003). In 2004, however, the tables were turned. Fifty percent οf the developers surveyed focused on Unix/Linux development, while only forty percent continued with Windows development. It seems that as business trends move more towards web-based technologies, the de facto standards for web services (currently Unix/Linux systems) have become the battlegrounds for market share gains. Developers are now turning their eyes towards a “new” arena, one that has been dominated by UNIX and Unix-like systems since the beginning. The conclusions that can be drawn from this OS market share analysis are simple. Microsoft Windows XP currently dominates the market in sheer numbers. Unix/Linux systems dominate the server market and are being developed now more than ever. The seemingly indestructible Windows platform has not monopolized the entire market, as many have been lead to believe. The battle for OS market share is just heating up. Hardware RequirementsIn general, UNIX hardware requirements are similar or slightly less than those οf Windows. One advantage οf UNIX over Windows is its scalability. Since required components can be added to UNIX as required, the operating system can remain as lean as possible. In contrast, Windows XP generally comes equipped with all required components whether one requires them or not. Windows XP Home Edition and Professional Edition have recommended requirements οf 300 MHz processor, with 128 MB οf RAM, and 1.5 GB οf hard disk space (Microsoft Help and Support Website, 2006 and Microsoft Windows XP Professional Website, 2006). Window’s server OS is Server 2003. Recommended hardware requirements are 550 MHz for general use and 733 MHz for data servers. RAM requirements are generally 256 MB, 1 GB for data servers. Hard disk space required is 1.25 to 1.5 GB (Microsoft Windows Server 2003 R2 Website, 2006). At minimum, a UNIX OS can run on a recommended 386 MHz with only 2 MB οf RAM. It requires 40 MB οf hard disk space, with an additional 2 MB if one also wishes to run X-Windows. X-Windows is a graphical user interface that simulates a Windows XP GUI, which the average user is familiar with and finds helpful. There are also cut-down versions οf UNIX that run on PDAs and only require 512kb RAM (Smart Soft Computing LTD., 2001-2005). Red Hat Linux 1.7 is recommended for use on 386 MHz with 64 MB οf RAM. 1.2 to 1.5 GB οf hard disk space will be required (Red Hat, Inc., n.d.). As a result οf this scalability, UNIX server requirements vary according to the usage environment. Some iterations οf UNIX based server OSs are Red Hat Linux 7.2, 8.0, and 9.0; SuSE LINUX 7.2, 7.3, 8.x; Solaris 7, 8, 9; HP-UX 11i. The Server Hardware requirements are listed in the table below. UNIX Server RequirementsDeveloperEditionStandardEditionEnterpriseEditionHardwareProcessorPentium for Linux, SPARC for Solaris, and PA-RISC version 1.1 or 2.0 for HP-UXRecommended RAM (MB)512512512Free hard disk space (MB)450 to install250 to run450 to install250 to run450 to install250 to runCapabilitiesRuns Web Servers such as ColdFusion MX, Apache Web Server, or Netscape ServerXxxRuns smaller scale Databases such as MySQL, Microsoft SQL ServerXxxRuns large scale Databases such as Oracle, Sybase Adaptive Server, or DB2 UDBx(Livedocs.macromedia.com, n.d.) In general, the home or workstation hardware requirements, in terms οf processor speed between UNIX and Windows, are generally similar. A 386 MHz processor will run both. Windows requires twice as much RAM as Linux, 128 MB verses 64 MB. Hard disk space is also similar, around 1.5 GB, although some pared down versions οf UNIX only require 40 MB. Server hardware requirements depend on usage . In general, RAM requirements are similar, around 256 to 512 MB. Processor requirements in UNIX based systems will run nicely on a 386 MHz processor, whereas Windows Server requires 550 MHz. There is an even bigger difference with hard disk space requirements. UNIX iterations require only 250 to 500 MB οf space verses 1.25 to 2 GB for Windows based. One reason for these differences is that in general Windows comes fully loaded with multiple systems, only a few οf which the average user actually requires. UNIX advocates might also point out that UNIX code is leaner and more efficient than Windows code. File Processing Understanding how an operating system processes all οf its data, especially critical files such as system or hidden files, is important since data processes are the life blood οf the operating system. Without data processes running, the operating system becomes useless. Although UNIX and Windows are two completely different operating systems, there are multiple similarities and differences between the two in regards to processing data. One similarity between UNIX and Windows is the way the file system is broken down. With UNIX, files are separated according to ordinary files, directories, special files, and pipes. Windows has a similar breakdown except that the files are named metafiles (system files), files or streams, and directories. The ordinary files in UNIX are the files that users work with and contain user information. The corresponding files in Windows with ordinary files are files or streams. The directories in UNIX and Windows serve the same purpose, which is to hold other files and directories. The pipe in UNIX is similar to temporary folders in Windows. Another similarity between the two operating systems is that they both allow blanks in filenames. However, it is not common to use blanks in UNIX for filenames since they are so confusing. A third similarity between UNIX and Windows is that they both support file compression. Windows has built-in support for file compression via NTFS while UNIX supports file compression through the use οf gzip. UNIX also has another utility called gunzip, which is used for file expansion. There are several differences between the two operating systems that set them apart from one another. The first is that Windows uses a backslash to separate directory names while UNIX uses a forward slash for directory names. Another difference is that UNIX must assign permissions by owner or group for executable files while Windows assigns file extensions, such as “.exe” or “.com”. A third difference between the operating systems, which is very noticeable, is that alphabetical letters are assigned to devices such as hard disks or floppy disks while UNIX mounts the device to a directory. What this means is that a floppy disk in Windows would be assigned an alphabetical letter while UNIX creates a filename similar to “/floppy/file1”. The last difference, which is also easily identified, is the format οf the commands used in each operating system. In Windows, commands such as erase, move, or copy are used. In UNIX, the same commands are cp, mv, and rm. Availability οf Application SoftwareAccording to Wikipedia, application software is defined as a subclass οf computer software that employs the capabilities οf a computer directly to a task that the user wishes to perform. Some οf the most common types οf application software are Word Processors like Microsoft Word, Spreadsheets like Microsoft Excel and media players like Apple Quicktime or Microsoft Media Player. Choice is one οf the most important things to consider when purchasing an Operating System (OS). Choice has been a mainstay οf open source systems like Linux/UNIX. Choice is the ability to choose any program and trust that it will install, run and not affect any other application (Devx, 2006). Application software should install with such ease that a user doesn’t have to know what GUI is being used. Also, a user should not have to make any changes to configuration files. Over the years each OS has had application software unique to its own system. To say that Windows applications far exceed UNIX/Linux would be an understatement. To find application software for Windows, one can simply go to a local retailer and find shelves full οf applications for Windows. Where as UNIX/Linux is an open source system and software is typically downloadable and free. Despite these differences application software for each OS is actually very similar. A good example οf these similarities is the spreadsheet programs used by Windows and UNIX/Linux. Perhaps the most well known spreadsheet program is Microsoft Excel. One οf its counterparts is Gnumeric. Both programs make use οf a Graphical User Interface (GUI). Both provide accounting functions, finance functions, statistical functions, and database functions (Linux, 2006). While both programs, on the surface, look similar, Gnumeric actually has 154 more functions than Excel. For example, the EURO function will convert Euros to any currency being used in Europe. For the mathematicians there is the ISPRIME function. This function will let the user know if the number entered is a prime number. As an added bonus, Gnumeric can import spreadsheets from other programs like Excel, Lotus and Quattro Pro (Gnome, 2006). Finally, and what is probably the best thing about Gnumeric, is that Gnumeric is free while Microsoft Excel 2003 can range in price from $73.00 to over $200.00 (Calibex, 2006). The main reason why Gnumeric is so much more robust than Excel is the fact that the source code is open and can be changed by any person or group. A program like Excel will let a user create formulas, macros, and even add plug-ins. However, a user is unable to change the source code and modify the program. This is because programs from Microsoft are proprietary, which means the source code is not available. This marks the principal difference between UNIX/Linux and Windows. User Interface User interface is defined as the aspects οf a computer system or program which can be seen by the user, and the commands and mechanisms the user uses to control its operation and input data (User interface, n.d.). There are various forms οf user interfaces that are currently being used throughout the computing industry, such as graphical user interfaces (GUI), command-line interfaces, batch interfaces, touch interfaces, and several others. UNIX/Linux and Windows XP, however, utilize specific user interfaces to allow users to interact with and control the O/S. UNIX/Linux uses two user interfaces, command-line interface and graphical user interface, to manipulate the O/S and all the installed programs on the computer system. Microsoft also utilizes the same two user interfaces, command-line interface and graphical user interface, for data and O/S manipulation. UNIX/Linux uses a system known as “X Window System (more commonly known as X11)” for its graphical user interface while Windows XP uses the standard windows that are commonly seen throughout all the Microsoft Windows releases. References Calibex.com (2006). Microsoft Excel – The Complete Package. Retrieved September 2, 2006, from http://www.calibex.com/Microsoft-Office-Excel-2003-79475574/cb2/prices-html Devx.com (2006). Linux vs. Windows: Choice vs. Usability. Retrieved August 28, 2006, from http://www.devx.com/opensource/Article/16969/0/page/3 Gnome.org (2006). Gnumeric – The Gnome Office Spreadsheet. Retrieved September 2, 2006, from http://www.gnome.org/projects/gnumeric/features.shtml Linux.org (2006). Application:HancomSheet 2.0. Retrieved September 2, 2006, from http://www.linux.org/apps/AppId_7750.html Livedocs.macromedia.com (n.d.). System requirements for UNIX. Retrieved August 19, 2006, from http://livedocs.macromedia.com/coldfusion/6.1/htmldocs/intro6.htm Microsoft Help and Support Website (2006). System requirements for Windows XP operating systems. Retrieved August 19, 2006, from http://support.microsoft.com/kb/314865/ Microsoft Windows Server 2003 R2 Website (2006). System Requirements. Retrieved August 19, 2006, from http://www.microsoft.com/windowsserver2003/evaluation/sysreqs/default.mspx Microsoft Windows XP Professional Website (2006). Windows XP Professional System Requirements. Retrieved August 19, 2006, from http://www.microsoft.com/windowsxp/pro/upgrading/sysreqs.mspx Petreley, N. (2003, March 14). Debunking the Linux-Windows market-share myth. Retrieved August 14, 2006, from http://linux.sys-con.com/read/32648.htmRed Hat, Inc. (n.d.). Hardware and Space Requirements. Retrieved August 19, 2006, from http://www.redhat.com/support/resources/faqs/rhl_general_faq/s1-install.html Refsnes Data (2006). Browser statistics. Retrieved August 13, 2006, from http://www.w3schools.com/browsers/browsers_stats.asp Smart Soft Computing LTD. (2001-2005). UNIX Tutorials -How To Make The Most οf UNIX. Retrieved August 19, 2006, from http://www.smart-soft.co.uk/unix/Unix.htmUser interface (n.d.). The Free On-line Dictionary οf Computing. Retrieved September 2, 2006, from http://dictionary.reference.com/search?q=user interface&x=38&y=10 Woodie, A. (2005, March 2). Windows continues to gobble up server market share. The Windows Observer, 2. Retrieved August 14, 2006, from http://www.itjungle.com/two/two030205-story03.html Notes [1] TeamQuest Optimizing IT Performance. URL Address http://www.teamquest.com/ html/gunther/. Updated July 30, 2003. Accessed August 17, 2003. [2] Gunther, Dr. Neil J. Solaris System Resource Manager: All I Ever Wanted Was My Unfair Advantage (And Why You Cant Get It). Computer Measurement Group Conference. Reno, Nevada. December 5-10, 1999. [3] Sun Microsystems. URL Address http://www.sun.com.au/products/software/ system_management/solaris_resource_manager_1.3.html. Updated August 6, 2003. Accessed August 17, 2003. Read More