Cloud Computing | Tutorial #28 | Google File System



hey guys welcome back to channel so into this video I will be discussing regarding what this Google file system in cloud computing so first of all let's begin with this so we have a common idea of how the Google works and you may be knowing like Google has so many products like Google Translate Google Drive you do and many other files and photos of different and billions of users so you might be thinking like Google will have a kind of different kind of specialized hardware or systems or some other kind of processes for handling this kind of data well that's not entirely correct so Google has got its own file system so that's called as a Google file system GFS so in today's video we'll be seen what does GFS and what are the various components and how the GFS are formed so let's begin so first of all let's focus on this diagram so you can see some components some blocks some small small figures and all this there so you have first of all this GFS master so this GFS master controls this client as well as this service so basically this GFS cross sells nothing but it has a GFS master then three chunks servers it has got and which serves five different lines so you have this GFS master and two clients which are there and these two clients are segregated into three different bifurcations and there are a number of chunk servers so this blocks are this dotted boxes which you can see are chunk servers and inside this you can see C 0 C 1 C 3 C 4 and C 5 so we have five different clients taken and so each of these clients are replicated into three different chunk servers like in case of any node failure happens so it's like in in some cases it's like if c0 fails to perform our c0 is in access building an octane c0 from this third replicar if you see what is not there then you can access it from the second replica so this is how it's structured so this setup like of a GFS master client and some kind of some set of servers are called as GFS cluster then this client servers and masters are all running on the Linux machines and each runs a server processes at user level and these are known as use a level server processes so all these have a kind of processes which runs at the server and it has operated the user level and so it's called as a user level server process next in this end a kind of setup that is a Google file system you have some kind of data that is data about data you have meta data so those meta data are all managed by the GFS master and it communicates effectively between the clients and the chong service so you have this setup so in this you know like normally in case we have the clients and servers now in this specially you have the chunk service and by the name chunk is there so chunks means a small block of data and it is opting from the file system so when the system files are created or when the configuration happens at that time we provide some kind of commands at runtime and then we create some kind of chunks or we just segregate into small small blocks and these are size of 64 MB so it's fixed size fixed way rate is there that is immutable and a 64 data bit is obtained in order to gather this chunk service so this chunk services are then replicated across the entire cluster and the clients interact directly with the chunk servers for data transfer so you can see this clients are then said this chunk server so this dotted boxes outside are the service so we have the three different replicas of the service and inside this we have small small clients are there so it communicates directly with the server so direct proportional communicate between the client and the server in GFS and chunks are replicated across the three machines in case of any node failure or reliability like I said if sees your fails to perform of seal zero is inaccessible then you can obtain at the third instance or the third cluster like if c5 is not there then you can obtain from the third one that is from here next GF estos three replicas by default and can be at any levels defined by the user so at first GFS has this setup like if you have a ZFS master and client and then server then it will automatically by default at the time of initiation we'll have three different set of servers but user has that controlled whether how many levels of application might have to designate so it's like entirely depending on the user whether he has to maintain the default three levels of replication or 5 or 6 or even at any higher number so likewise the size of that cluster will go on increasing then the system interface is same like any other system like normal you have the user interface here the command line the graphical user interface and all like that and especially provides snapshot and happen to operation kind of system interface and this specially allows to create a copy of the files or the folder system in a more cheaper way than by any other means and the applications contain specific file system that is FS AP is executed by the code like for some client application you have some client code to have be written at that place so it needs to be executed at some time or a later time and it's communication is done by the gfs master and the chunk service is established so first of all whenever any client has to be communicated with the gfs master and the circle then uh code has to be written for the client and then it establishes the communication link or a network will be there that links are this one so it communicates with the gfs master and the chunks occurs then in most of the file systems you encounter a POSIX API where as in gf this POSIX API is not included and so any need of Linux Vinodh layer cooking is not required in specially in case of GFS like in any other file system apart from GFS you can see that POSIX API and one four most important thing is that no caching of the file data is performed by the clients or the servers because because in Google file system you have all the data in a streamlined workload so it's like in a stream you get in a particular set of windows inserting certain data links or data bits is there so because of that streamline workload or stream workload caching does not benefits the clients by any means but whereas to some extent service has at least a consequence of buffer cache already which maintains a record of this frequently requested local files or frequently accessed local files so there is no need for any caching because service will have its own buffer where as clients it need not have any kind of access for frequent local files so that was all regarding the Google file system and cloud computing so hope you guys enjoyed this video if you want a ticket by watching this video please do like share comment and most importantly don't forget subscribe to my channel thank you very much for watching this video

6 Comments

  1. vaishna shetty said:

    ♥️🙌

    June 26, 2019
    Reply
  2. Tabu Sk said:

    Hindi mai samjao plzzz

    June 26, 2019
    Reply
  3. shad khan said:

    wo sb to thik hai but ittaa jump q kra starting me..???

    June 26, 2019
    Reply
  4. Datta Narhare said:

    hindi me samjate toh acha hota…

    June 26, 2019
    Reply
  5. siddhant shinde said:

    nice

    June 26, 2019
    Reply
  6. Rajat Badade said:

    Please prepare these videos in Hindi

    June 26, 2019
    Reply

Leave a Reply

Your email address will not be published. Required fields are marked *