Locks In Disributed Systems
- 1. Locks In Distributed Systems Mridul K Mishra(170303201015)
- 2. What is it? In computer science, a lock is a synchronization mechanism for enforcing limits on access to a resource in an environment where there are many threads of execution. A lock is designed to enforce a mutual exclusion concurrency control policy. In Distributed Systems(referred to as DS from now on), lock is a mechanism that allows only one of the innumerable nodes(or process) to access and modify a resource or data that is being shared commonly to prevent execution of same task twice and also maintain data integrity. In the common database of a bank, at a given instance of time only one node(at bank branch or user himself using personal banking) should be able to carry out a transaction on a given bank account ( Data Integrity).
- 3. Why is it needed? Any DS always uses a database, and concurrent tasks are always performed by different nodes, so analogous to the relational database, R-W anomaly W-R anomaly W-W anomaly may occur. This ensures integrity and consistency of data. This lock is for Correctness. In addition to this, the system has to ensure that only one process is fired for execution of a single task, e.g. in an inventory, upon shortage of a needed resource, order should be placed only once. This increases the efficiency and throughput of the system. And similar to single computer OS, problems like Deadlock can occur, which have to be prevented. This lock is for Efficiency.
- 4. The basic idea behind locks is to decrease the cost of carrying out operations in a DS by ensuring that among several nodes trying to do a task only one does it(at a given instance of time). Although locks are valid in DS for both above stated scenarios, the locking mechanism should be chosen wisely according to the need. E.g. In many cases applying the lock to prevent repetition of task might be costlier than task itself.
- 5. How Is It Implemented? Compared to conventional OS and relational database, the locking mechanism in a DS is quite complicated as the network itself or the nodes can independently fail at any time. There are various algorithms deployed for this in different DS, E.g. Redlock, Paxos. The lock managers are generally referred to as Distributed Lock Managers(DLM) The DLM uses a generalized concept of a resource, which is some entity to which shared access must be controlled. This can relate to a file, a record, an area of shared memory, or anything else that the application designer chooses. A hierarchy of resources may be defined, so that a number of levels of locking can be implemented. For instance, a hypothetical database might define a resource hierarchy as follows: Database Table Record Field A process can then acquire locks on the database as a whole, and then on particular parts of the database. A lock must be obtained on a parent resource before a subordinate resource can be locked.
- 6. Various types of Locks that are granted by DLM in order of their hierarchy are: Null(NL):Process wants to lock resource, but does not stop others from getting lock on it. Concurrent Read(CR):Multiple processes can read a resource but not update it. Concurrent Write(CW):Multiple processes can read and update a resource but no exclusive access provide. Protected Read(PR):A process( or multiple) reads a resource but prevents others from updating it Protected Write(PW):Indicates that a process will read and update a resource. Prevents others from updating it. CW lock holders can read it. Exclusive(EX):This is the traditional exclusive lock which allows read and update access to the resource, and prevents others from having any access to it
- 7. A process can upgrade its lock as per need and permission of DLM. The following table explains the relation between above stated locks:
- 8. A lock value block is associated with each resource. This can be read by any process that has obtained a lock on the resource (other than a null lock) and can be updated by a process that has obtained a protected update or exclusive lock on it. And based on above tables locks can be updated.
- 9. Issues Although DLMs provide for data integrity some issues that might arise are: Lost Update: say we have an application in which a client needs to update a file in shared storage. A client first acquires the lock, then reads the file, makes some changes, writes the modified file back, and finally releases the lock and in meantime the lock/lease expires. Another client acquires the lock The lock prevents two clients from performing this read-modify-write cycle concurrently, which would result in lost updates. Explained by diagram.
- 10. Using Timestamps/tokens can solve this issue. Deadlock: When one or more processes have obtained locks on resources, it is possible to produce a situation where each is preventing another from obtaining a lock, and none of them can proceed. This is known as a deadlock. A simple example is when Process 1 has obtained an exclusive lock on Resource A, and Process 2 has obtained an exclusive lock on Resource B. If Process 1 then tries to lock Resource B, it will have to wait for Process 2 to release it. But if Process 2 then tries to lock Resource A, both processes will wait forever for each other.
- 11. Advantages: Ensures the Correctness of transactions. System becomes more robust i.e. prevents system failures Illustrations: Chubby- Designed by Google for loosely coupled DS Apache ZooKeeper- An open source manager designed by Yahoo. Redis - an open source, BSD licensed, advanced key-value cache and store
- 13. Thank You