Memory Management In Distributed Operating Systems Computer Science Essay

Rainbow OS [ 1 ] is a transactional distributed memory operating system for PC-clusters, designed for 64-Bit multicore architecture. Rainbow OS is wholly implemented in Java and compiled into native codification with a particular compiler [ 4 ] . Therefore, Rainbow OS offers non merely system but besides application programmers the advantages of a type safe linguistic communication at the same time giving them a fast runtime public presentation. The design of Rainbow OS provides two memory direction grounds which are ; for the local memory direction that is integrated in the so called local meat, being responsible for the lowest address part of the 64-Bit reference infinite and the execution of several basic system maps.

Rainbow OS offers the usage of entirely useable local meat for each node of the bunch and plus, it is local. The design and functionality of the locally memory direction of the local meat will be illustrated in the following chapter. Quite the antonym to the local meat, the alleged distributed meat of Rainbow OS is in charge of the higher 64-Bit reference infinite which includes extra shared system maps.

Get quality help now
Verified writer
5 (339)

“ KarrieWrites did such a phenomenal job on this assignment! He completed it prior to its deadline and was thorough and informative. ”

+84 relevant experts are online
Hire writer

This meat is wholly distributed and shared by all nodes of the bunch and some of the challenges are to keep the consistence and the shared memory direction for right functionality of the bunch with its nodes. To do certain of these demands, Rainbow OS implements a transactional consistence mechanisms. Additionally, the Transactional Consistency Rainbow OS besides provides other consistence theoretical accounts so that application coders can pull strings and command their ain consistences that are suited with their demands.

Get to Know The Price Estimate For Your Paper
Number of pages
Email Invalid email

By clicking “Check Writers’ Offers”, you agree to our terms of service and privacy policy. We’ll occasionally send you promo and account related email

"You must agree to out terms of services and privacy policy"
Write my paper

You won’t be charged yet!

Primitive information types ( whole numbers, longs and etc ) are used as the chief informations that can merely be used by other consistence theoretical accounts to protect Rainbow OS and its type system. Not merely that, Rainbow OS uses a particular object design that is different to the objects of Sun Java to supply the several consistence theoretical accounts. Further inside informations are made known in the following chapters of this paper.


The distribution of all informations in the bunch every bit far as possible is one of the challenges faced in planing Rainbow OS. Therefore, simple object constructions are non the lone thing shared by Rainbow OS but besides codification that is useable by all nodes. Because of the difference in system critical constituents and hardware features, they have resulted in bunch that can non be shared if all the informations are to be put in the same bunch. At the start for every node, installations for the boot procedure or for web communicating must be entirely available if it exists, before it is able to fall in the bunch. In supplying the best runtime public presentation, system or any hardware specific constituents such as web arranger which is bounded statistically are implemented in the local meat of Rainbow OS. Following the startup procedure, all necessary drivers and little subset of system tools are initialized by local meats before the node can fall in the bunch. Because of drivers and memory direction of the local meats are sole for every node and non shared in the bunch, every node and its local meats are comparable with a individual station system. On the contrary, Rainbow OS & A ; acirc ; ˆ™s distributed meat is wholly distributed compared to the local 1s and therefore, shared by all nodes. The distributed meat will be loaded from the shared memory of the bunch right after the boot procedure and initializing of the local meat. Bunch can be loaded from a particular waiter ( alleged boot waiter ) or a separate storage medium such as CD-ROM, if there is none. The deduction of one shared memory direction for all nodes in the bunch is done by the distribution of the distributed meat. Therefore, the position of the shared memory that is the individual system image, and the use of the same codification and informations constructions of the distributed meats are being shared by every node. An overview of the memory direction of local and distributed meat will be given in the following two paragraphs. Furthermore, the distributed meat and its informations are under control of Transactional Consistency to do certain of the guarantee of right functionality of the bunch, particularly of the distributed memory direction ( see paragraph II.D ) .

Figure 1: Startup procedure of Rainbow OS

Local Memory Management

For Rainbow OS, the local memory direction topographic point the care of physical pages of each node, page tabular arraies and besides offers system or driver coders memory allotment of two memory pools. These pools are placed in between the distributed meat and the local one ( mention paragraph II.C ) . Java is the good known new interface dedicated to coders to apportion memory. In contrast to Sun Java, another internal construction of its objects is used in Rainbow OS. Primitive information types ( scalars ) and mentions are objects that are separated in Rainbow OS. Mentions to objects that separate these two parts are stored below and scalars above ( figure 3 ) . The chance to hive away scalars at another location in memory is a specific characteristic of Rainbow OS outside of the proper object. Best public presentation and a fast startup procedure are offered through these alleged indirect scalars that are non used in the local memory direction which keeps it little and avoiding unneeded complexness. Information on the construction and use of indirect scalars are shown in paragraph II.C. Creation and the disposal of page tabular arraies for the logical reference infinite are other maps of Rainbow OS. It is inappropriate and impossible to make and hive away all page tabular arraies for the full 64-Bit reference infinite with the current capacity of memory bunch nodes utilizing a 64-Bit architecture. Therefore, the local memory direction supply an on demand allotment and creative activity of new page tabular arraies which keeps the set of page tabular arraies smaller at startup. Other than that, if the matching reference infinite is fresh, parts of page tabular arraies can be removed to cleanup memory. The memory direction militias at startup a specific logical reference infinite to map and entree all page tabular arraies, due to the allotment and creative activity on demand that has been mentioned before.


A 64-Bit architecture multicore processors is supported by Rainbow OS where at present, there is merely one nucleus that is able to fall in the bunch and works genuinely distributed where the other are merely locally available at the minute and this, is our chief end which is to back up each nucleus of a multicore processor in this manner. Consequently, independent node in the bunch is represented by each nucleus of a processor in each bunch and to enable multicore working in bunch, a few demands should be met. In order to vouch a correct processing, hardware constituents that are alone in every bunch Personal computer such as physical memory or web arranger, need to be synchronized with all nucleuss of one processor. As an illustration, to avoid duplicated allotment of physical pages, physical memory direction has to be synchronized or partitioned.

Figure 2: Local memory direction

Figure 3: Object construction

Not merely that, page tabular arraies can be differed and can non wholly be shared if each nucleus could utilize other subdivision of subdivision memory with respects to the mechanism of Transactional Consistency. Further research will concentrate on the complete support of multicore central processing unit mentioned before.

Distributed Memory Management

Rainbow OS possess a local and a distributed memory direction as mentioned antecedently and the latter is implemented in the distributed meat and wholly distributed in the bunch. An extra communicating installation between both meats is necessary since two meats exist in the system which are the local and the distributed and normally, it is fact that non every driver or system constituent can be shared by all nodes. That is why Integer-Interface is used by Rainbow OS with ain informations buffers to pass on between the local and distributed meat widening the range in which codification can be shared by nodes of the bunch. To exemplify, the distributed meat of Rainbow OS can implement drivers excessively. Besides, several consistence theoretical accounts can be supported by Rainbow OS for its shared informations and for this intent ; memory parts are the consequence of logical reference infinite division. Each part is under control of one consistence theoretical account which in size of 512GB therefore ; the upper 25 reference spots specify the consistence which is the same with the upper hierarchal degree of page tabular arraies. ( Hardware inside informations are shown in chapter IV ) . A fast categorization of the used consistence at runtime is offered by this design. Transactional Consistency is used by Rainbow OS as default consistence which covers the distributed meat, the shared system maps and all user application and their informations.

Figure 4: Distributed memory direction

The allotment of memory and object creative activity in their corresponding memory part is managed by the alleged distributors. They are objects themselves crossing certain address part inside of a memory part and are subdivided into two parts ( see figure 5 ) . One portion has new objects with their several mentions every bit good as direct scalars and the other has all indirect scalars. As a consequence, the capacity of the selected distributor is reduced by each memory allotment until its capacity is exhausted or the distributor object is cleared in memory. To vouch the safety of the type system, all object mentions are usually under the control of Transactional Consistency therefore, every distributor has to be ever created in a memory part and being under its control. Second, the needful consistence theoretical account is able to utilize by the coders which Rainbow OS features the chance for coders to utilize weaker consistence theoretical account for noncritical informations with no cause danger to the system. How we can loosen up this demand without seting at hazard for the runtime system is our future research & A ; acirc ; ˆ™s concern. Programmers of application or system tools can do their ain distributor avoiding latent hits amid different applications and their objects for sole memory use ( e.g. false sharing ) .

Figure 5: Allocator object construction

Transactional Consistency

In Rainbow OS, all calculations are enscapulated in minutess if to be compared with the well recognized procedures of other runing system. Transaction consequences are written to memory and the original province of a page is preserved by a shadow transcript. All shadow transcripts are restored and the dealing will be restarted automatically if in instance of an abort. The references of their written pages are sent to all nodes of the bunch at the terminal of a dealing. Therefore, the nodes can observe hits between viing minutess. A coincident dealing has to be restarted if it has read one or more modified pages. The local memory direction will observe all of the set pages which are read or written which requires fringy package action. No locks are needed to entree memory or shared objects for Rainbow OS during executings because it uses an optimistic synchronism, leting minutess to continue locally and its proof at the terminal of it. [ 2 ] [ 6 ] [ 7 ]


Several consistence theoretical accounts are supported by Rainbow OS which kernel or system coders are the implementers and each consistence covers a separate logical reference infinite which has a size of 512GB. Rainbow OS has per default three consistence theoretical accounts: a local, transactional and a particular one for memory mapped IO which can be used by device drivers to map their device memory. The preset for all mentions in Rainbow OS is Transactional Consistency where it inhibits invalid uses of the meat runtime constructions of Rainbow OS or the type system of it. Additionally, weaker consistence theoretical accounts for mentions will be examined by us without seting maps at hazard. Rainbow OS is watchful of the entree of every memory and delegates it to the several consistence protocol where it can work harmonizing to its demands. This will do possible of the execution of new consistence theoretical accounts in Rainbow OS without complex versions in system maps.


A bunch node is used for the measuring, configured as follow:

Processor AMD Athlon ( thulium ) 64 X2 Dual Core Processor 4200+

Cpu MHz. 2199.754

Motherboard: MSI K8T NEO2 V2.0

Memory: 1024 MB DDR2-RAM

One of the chief maps of every memory direction is memory allotment in the position of Rainbow OS, memory allotment and object creative activity are included excessively. Therefore, the clip to apportion memory and the creative activity of new objects have been measured by us including all processs for a new ( ) in Java. The same processs implemented in Sun Java on the same hardware have been compared with the consequences that we have. A choice of the consequences that we have gathered naming the clip to make a simple object is showed in the undermentioned overview which contains no extra Fieldss or mentions:

Table 1: Measurements of memory allotment and object creative activity

In Sun Java, The Measurement of a new ( ) includes automatic refuse aggregation and etc as extra functionality inside the Java Virtual Machine. On the other manus, farther functionality with respect to distribution and consistence is included in our object creative activity and memory allotment. In add-on, the potency for our memory direction in Rainbow OS is some optimisations where it will cut down the mean clip of a new ( ) .


An efficient execution of a refuse aggregation and working analogue over all nodes in the bunch are two of our following challenges therefore, every creative activity of objects with its mention in the harmonizing object form have to be noted by the memory direction which is portion of the system information. Therefore, a mention will be contained by all forms to a particular container with a particular consistence that is, the alleged relaxed back packs which mentions of all cases of a certain object are included. Objects are traceable in memory as the footing of this execution. Garbage aggregations could be implemented in several ways and one of the ways is really uses the type of information of Rainbow OS where dealingss and dependences for an inspected object are to be identified. Hence, the possibilities that we have to analyse all sorts of mentions to any object. It is determinable whether an object is refuse or non with sort of information and another possibility to place refuse is simple by utilizing an offline refuse aggregation which a particular cluster-PC ( alleged page waiter [ 3 ] ) is implemented in it. Owing to the possible hits during the entree to the relaxed back packs, it will be stored in a specific consistence theoretical account conforming to a reduced semantic that guarantees each node the same content but in perchance different order.

VI. Position

To better an expeditiously working refuse aggregation for shared memory, several consistence theoretical accounts for Rainbow OS will be designed and implemented, back uping system and application coders and their executions to accomplish best runtime public presentation. The strongest consistence theoretical account in Rainbow OS is Transactional Consistency which guarantees the type system & A ; acirc ; ˆ™s safety therefore no 1 can interrupt it ( in malicious purpose or unintended ) and do danger to the runtime system. Weaker consistence theoretical accounts are available for this ground for crude informations. The impact on the public presentation when utilizing weaker consistence theoretical account will farther be examined for indirect scalars of objects. Furthermore, to loosen up the consistence limitations for mentions can be advantageous for some instances. When mentions stored in a consistence country with a weaker consistence as Transactional Consistency, it could be utile. On the other manus, Rainbow OS type system has to be protected and therefore, the focal point is on how we could implement weaker consistence for mentions and in which instances it is utile and noncritical.

Cite this page

Memory Management In Distributed Operating Systems Computer Science Essay. (2020, Jun 02). Retrieved from

Memory Management In Distributed Operating Systems Computer Science Essay

👋 Hi! I’m your smart assistant Amy!

Don’t know where to start? Type your requirements and I’ll connect you to an academic expert within 3 minutes.

get help with your assignment