23. Two Phase Locking (2PL) | System Design

[@shamstabrez2986]

kya admi ho bhai u r covering all the important system design concepts keep going keep doing sch kind an amazing work we r here to support u

[@ConceptandCoding]

🙏

[@saurabhtiwari9614]

Kl hi DBMS revise kra tha aaj aapka lecture deekha. Kasam se aatma tript ho gyi.🎉🎉🎉 Ek request or h sir. Agar aap small small practiceal demo ki video bhi provide krwa de toh mza aa jayegaa. 😊😊😊

[@ConceptandCoding]

:)

[@HimanshuKumar-xz5tk]

This is perfect explanation.

[@manoharmanu9240]

Great explanation👏👏

[@mayankkumarprajapati2337]

again , very good video, really liked the detailed explanation

[@ConceptandCoding]

Thanks

[@shubhamagarwal1434]

very well explained.......awsome!!!

[@priyanshnamdeo9504]

In the 35:04 example ; since we are dealing with Basic 2PL ; txn 1 should acquire locks on both A and B during the growing phase , that way there shouldn't be an issue , Correct me if I am wrong ; the example that you have have cited for Basic 2PL seem to be violating the Basic 2PL principle .

[@pavankumar-cy6mg]

it is not breaking the basic 2pl principle, as the name suggest it is a growing phase it will aquire lock only when needed, as for A it acquried first and for B later, if it would have acquired both locks at the begining then it would have became a conservative 2PL isn't it?

[@girishanker3796]

​@@pavankumar-cy6mgat 6:10 he is saying the graph can't be like what you are saying, like lock- unlock - lock. So whatever Priyansh is saying is right.

[@baburaomulaparthi4531]

explanation is very good bro, can you share programming examples.

[@garimadhanania1853]

One question: The last example that you take at 35:00 for basic 2PL, did you come up with or if not, could you please share the source of it? Its odd that you one would update A (commit) before crediting to B. I assumed that credit debit kind of transactions happen in an atomic manner. In basic 2PL the locks could be release before the commit, but when the commit happens both A and B's value would be changed. Thats how I imagine correct "atomic transaction" to be implemented

[@akhilnama1989]

Hi really liked your content, it will be very helpful if you can please make notes available for other videos also. Thanks

[@ConceptandCoding]

ack

[@AkashYadav-ij1we]

Good theory, it would be better if we could have also implemented each of them. Just a simple demo.

[@ConceptandCoding]

Sure I will plan something for it

[@anshulkatare]

If transactions can release a lock on a resouce as per their wish, then its not really a Repeatable Read isolation level or higher. Which says a txn will keep shared lock or Exclusive lock till the end of txn. How does isolation level and unlock at will work together?

[@ApoorvBansal-s7d]

Hi sir, in case of Basic 2 Phase locking can we prevent deadlock at code level also by making request serializable and access of critical section by using semaphore and mutexes.

[@srijandasbiswas4864]

Can you please make a video on jwt token, OAuth authentication . It will be extremely helpful

[@ConceptandCoding]

Noted

[@dhanashripatkar5661]

Hi Shreyash, thanks for spreading this knowledge in a simple and easier way. One request for you, can you please create a video on Docker. Many times conceptual questions based on Docker were asked me in an interview. Also, in one system design interview, I was asked to design an "Online Code compiler System which supports many programming languages". How to do this with docker implementation?

[@ConceptandCoding]

Ack, i will plan for it in HLD series. Thanks a lot for the information Dhanashri

[@prateekkanujiya9775]

Hi Shreyansh, Could you please suggest how can we achieve 2PL using Java ?

[@Dushyant-j2e]

2PL will be achieved on DB level not on application level. Application level locks can only prevent other threads from concurrency but in a distributed system multiple processes would try to acquire lock which can only be handled at DB level.

[@SaiPhaniRam]

Timestamp 27:56 If TXN 1 has Higher Priority compared to TXN 2, then how come TXN 2 Exclusively locks the record A earlier ? The very point of precedence is not served right?

[@manojmrpd]

Good Video

[@ashishchoksi8501]

Good explanation!

[@ConceptandCoding]

Thank

[@ashishchoksi8501]

@@ConceptandCoding Hi, I'm struggling to understand exact difference between sharding and partitioning. can you please link me to the video that explains well? or please make one 🙏

[@ConceptandCoding]

@@ashishchoksi8501 noted buddy, i will cover the in depth video of it

[@manthankhorwal9531]

@@ConceptandCoding Sharding is a type of partitioning only.

[@anadilfaridy]

I see only 2 deadlock prevention strategy timeout and graph but in video you said you explained all 4, am i missing something?

[@girishanker3796]

3. Conservative 2 PL 4.TimeStamp based Deadlock detection

[@ShashikantPawar-g5u]

Can you please complete this series as per roadmap in first video

[@ConceptandCoding]

Yes working on it

[@prashantkumar-hx1dv]

Where is the link for getting the membership of "channel's members on level: Unlock LLD, HLD and JAVA (or any higher level)." Please tell. I have only limited access.

[@zainulabidin5133]

Hi Shrayansh, Cascading abort as per my understanding is the dirty read problem, in your previous video you told that pessimistic locking is done for isolation levels "Repeatable Read" and "Serialization" in which dirty read can not be done. Can you kindly clear this please?

[@mohammadkaif8143]

Yep, thats why its the job of the scheduler to abort such transaction( to avoid dirty read) and to fullfil the contract

[@saurabhmittal6947]

in the example shown at 30:40, I assume, we are using Repeatable read isolation level then how is txn1 is able to release the write lock in middle ? shouldn't it release at after commit ??

[@punitrai9338]

I am also having the same doubt, did you got the answer from somewhere ?

[@akshaysachan9592]

​@@punitrai9338 @saurabhmittal6947 I believe, he mentioned in the last video that Unlock operation can be done manually even before commit. Be default if commit happens, unlock is also done, but unlock can be done explicitly as well, and that is where Transaction2 put down X(A) lock.

[@garimadhanania1853]

thank you!

[@ShealImon]

Hi Shreyansh, I saw you at Ghee Factory 😊

[@ConceptandCoding]

ghee factory?

[@ShealImon]

@@ConceptandCoding you bought cow ghee!

[@ConceptandCoding]

@@ShealImon :) which video you have seen buddy

[@ShealImon]

@@ConceptandCoding I was just watching random videos, Ghee factory in Jaipur. Suddenly I saw you, It was 5 months old video.

[@ConceptandCoding]

:) yes gavyadhara ghee process, i remember now.

[@clutchh_god]

in basic 2PL example at 35:04 , since all locks are first acquired gradually and after then only it gets released gradually. How can we unlock A before acquiring B?

[@ConceptandCoding]

If txn wants B, then it can not unlock A else it will break the 2PL (growing phase) and shrinking phase concept

[@clutchh_god]

@@ConceptandCoding in that example txn requires B as well so ideally it should not unlock A before right?

[@ConceptandCoding]

@@clutchh_god yes

[@rajshah9129]

@@ConceptandCodingif txn1 wants B then it will put X(B) so txn2 will not able to put S(B) untill txn1 has commited and after that txn2 will read right data so i don't find any problem can u explain what issue their??

[@ConceptandCoding]

@@rajshah9129 could you please elaborate usecase a bit more, i could not able to get the complete question

[@clutchh_god]

since 2PL is a type of pessimistic then how is cascading failure possible, since isolation level would never be read uncommited?

[@ConceptandCoding]

Generally we call 2PL as type of pessimistic locking bcoz: - it also acquire lock But it doesn't follow exactly the same as pessimistic isolation strategy. Strict 2PL is very similar to pessimistic lock where locks are released at the end.

[@clutchh_god]

@@ConceptandCoding but locks are acquired in OCC as well

[@ConceptandCoding]

@@clutchh_god no locks is not required at OCC. The lock before update can be avoided in OCC by putting one Row version condition in where clause

[@ConceptandCoding]

Check the comment section of Concurrency video there is discussion on that.

[@harshagarwal_net]

@@ConceptandCoding update itself apply lock on rows that lock. will happen right?

[@nishantkumar6116]

2PL is a type of pessimistic locking, and pesimistic locking have isolation level of types Read Repeatable and Serialized and in these locks are accquired till the end of the transaction , then how we are taking the examples where 2nd transaction is able to accquire lock before 1st transaction getting committed, Can someone please help me with this

[@ConceptandCoding]

2PL is called type of pessimistic locking because of it nature of acquiring lock on the resources. As pessimistic also put lock. You can say Strong strict 2PL is more LIKE pessimistic where it releave the locks at the end of the txn only.

[@nishantkumar6116]

@@ConceptandCoding got it thanks!

[@akshaymahajan9626]

Great.

[@kochu7484]

Was here only :p Anyways if seeing this, when is spring planned any updates buddy?

[@ConceptandCoding]

Yes planning to do some strech this month and completing Java. And next i will start spring

[@AliHaider-gp1rb]

I am new to this topic, may be I am wrong but correct me if I am I have a question here that in the last section where you are explaining the example of basic 2PL when the shrinking phase start for transaction 1 at time sequence t4 i.e. unlock the data A the transaction has to be commit or wait according to the definition of 2PL because the shrinking phase started and after shrinking phase transaction cannot start the growing phase again as its doing on time sequence t11 where it acquiring the lock on B. No offense again saying I am new to this please correct me if I am wrong but as per my understanding until now transaction 1 can never acquire any locks once it started its shrinking phase.

[@BeingRahulDubey]

having same doubt

[@ujwalkumar2374]

That unlock(A) is not correct. It should not be there. If you ignore that, then T1 and T2 has taken a shared lock on A. Post that, T2 took exclusive lock on B, committed and released the locks. So no lock is released before the commit.

[@anantsingh2004]

Hi shrayansh, bhai your 2 videos are unavailable in this playlist ex - hld of url shortening service, where/how can I watch it ?

[@ankitsh6557]

Information is good. But try to explain clearly. And avoid scrolling a lot. Wait die and wound wait concept is not explained properly.

[@ConceptandCoding]

Sure Ankit thanks for feedback, will make sure of the scrolling part in future

[@billion_dreams__7787]

Subscription liya hu but chal vídeo chal nahi raha h

[@ConceptandCoding]

i see you have joined as "appreciation level", for lld/hld videos, buddy you need "Unlock LLD, HLD, Java" level membership.

[@billion_dreams__7787]

Java vala playlist open hi Nhi ho raha h

[@billion_dreams__7787]

@@ConceptandCoding bhaiya kevel main java ke subscribe kiya tha vo bhi nahi open ho raha h

[@ConceptandCoding]

Bcoz abhi "appreciation level" hi show go Raha hai. Ho sakta hai thodi der mein update ho jaye buddy

[@billion_dreams__7787]

@@ConceptandCoding bhaiya aap apne side se java ka playlist 1 month ke liye karva dena

[@harshitagarwal2682]

👍👍

[@mohammednizamuddints7736]

explanation is very good bro, can you share programming examples

[@apoorvbansal2179]

Hi sir, in case of Basic 2 Phase locking can we prevent deadlock at code level also by making request serializable and access of critical section by using semaphore and mutexes.