DDIA: Chp 9. Consistency and Consensus

-

The chapter 9 of the Designing Data Intensive Applications (DDIA) book has the following 4 sections (which contain a total of 10 subsections).  

  • Consistency guarantees
  • Linearizability
  • Ordering guarantees
  • Distributed transactions and consensus

TMI (Too much info)

The chapter tries to do too much. Almost an entire semester of distributed systems content is force-jammed into this chapter. I don't think the discord reading group will be able to complete this chapter and make sense of what is going on. For a more manageable load, much of this chapter should be moved to other chapters.

The chapter starts with explaining linearizability. It would have been better to explain this in Chapter 5, Replication. Linearizability sounds easy, but it takes time to internalize. I had discussed linearizability here, and later revisited it with this paper. And of course it was mentioned in many other posts.

The chapter continues with CAP tradeoff, rightly calling it "the unhelpful CAP theorem", but then since this is a crowded chapter, there isn't enough space to talk about what was wrong with CAP's misinterpretations and misuses, and what do we have in our arsenal to attack these problems.  I talked about CAP here and here, and mentioned it in several places as well. 

Then the chapter goes onto "ordering and causality". That is another huge topic that should have been its own chapter. It takes time (pun intended) to understand the happened before, causality, and concurrency concepts. In addition to Lamport clocks, this discussion should also include vector clocks. As the book says, linearizability is not the only option out there. In order to build such applications/algorithms,  we should first build the abstractions and the tools to capture partial orderings.

I really liked the Walter paper (SOSP 2011) as it managed to bridge the two worlds: 2PC based reconciliation of the updates with CRDT based no-coordination updates. Walter uses multi-version concurrency control within each site, provides a global causal order for regular objects (either doing the update at the owner site or using 2PC to coordinate across sites), and is also able to avoid coordination for certain tasks by using CRDT objects exclusively. In sum, Walter (using Parallel Snapshot Isolation) ensures consistent snapshots and transaction ordering within a site, causal ordering across sites, and prevention of write-write conflicts globally.


State Machine Replication (SMR) and consensus

For furthering the discussion of linearizable distributed systems, I don't think introducing the total order (atomic) broadcast is helpful. I think state machine replication (SMR) is a much better abstraction to think about this concept. The Tango and Delos papers explains things in terms of SMR.

Chapter 5 introduced replication for storage. I think it would be better to introduce Raft/Paxos there, and then also show how to move from atomic storage problem to the SMR maintenance problem as a more general solution. I think the new version of the book should introduce how Paxos/Raft protocols work as most distributed systems had to include/customize them as a component.

Consensus is easy to solve if you don't care about fault-tolerance: just have a single leader responsible. But leader failover when that leader crashes, or worse becomes unavailable for sometime for some of the system, is very challenging and leads to all kinds of cornercases including split brain scenarios. It is easy to underestimate how difficult a problem this is, and easy to mess up things as you cook up  your solution or even modify/customize an existing Paxos/Raft implementation. So that is why it is important to cover Paxos/Raft.

As for this chapter, Chapter 9, to keep things concrete it would be nice to show how a Paxos-groups based approach (like Spanner, MongoDB, CockroachDB) provides SMR and cross replicaset transactions with 2PC running on top of Paxos-maintained replicasets. This pattern is the most popular approach to building scalable distributed systems in production. 


Atomic commit

The chapter introduced the atomic commit problem which is a related problem to the consensus problem. In the consensus problem: multiple nodes must agree on a single value, any proposed value can be chosen, and we can make progress if majority agrees. In the atomic commit problem: multiple nodes must unanimously agree to a commit or abort decision (often for a transaction), the only two possible values are  commit or abort, and majority participation is not enough (since a single "no" vote forces abort, we need to account for everyone).

2PC protocol (which I summarized here) provides a simple solution for atomic commit. Well, a simple but fault-intolerant solution... Since you need fault-tolerance, this simple solution is not acceptable (analogous to just using a single leader is not acceptable for solving consensus). As I mentioned above a common approach is to run 2PC over Paxos groups so each participant is virtually infallible, and using 2PC would work. Or you go with a fault-tolerant SMR solution like Tango Delos that leverages a log for the entire system.

Other solutions are also possible. Since the two problems are related, there is interesting bridging opportunities. The U2PC paper shows how atomic commit and fault-tolerant consensus reconciles as special cases. But again, understanding this paper requires a lot of background in consensus, Paxos, and even Fast Paxos, to appreciate that bridging in the first place.


A nitpick

The book had established isolation properties in Chapter 7: Transactions, so I was confused why it repeats here (in this most crowded section) that linearizability is not equal to serializability. Linearizability is a consistency property and Serializability is a transaction isolation property.

I also have a beef about the book calling Serializable Snapshot Isolation (SSI) not linearizable. 

However, serializable snapshot isolation (see “Serializable Snapshot Isolation (SSI)”) is not linearizable: by design, it makes reads from a consistent snapshot, to avoid lock contention between readers and writers. The whole point of a consistent snapshot is that it does not include writes that are more recent than the snapshot, and thus reads from the snapshot are not linearizable.

I think SSI (summarized here) does not preclude linearizability per key. That a transaction does snapshot-isolation reads for its readset from a transaction-start point does not make SSI violate linearizability. Reads as part of the transaction are subject to transaction jurisdiction property of isolation, in which case we have SSI as the isolation property, meaning those transactions are serializable with respect to other transactions. It would make sense to say that SSI does not guarantee strict-serializability (SSER) (a clock order), and maybe the book means to say that. 

In fact a better example to provide here would be CockroachDB. CockroachDB provides linearizable reads/writes and serializable transactions, but it still cannot provide SSER transactions, as I explained here.

Comments

Post a Comment

Popular posts from this blog

Hints for Distributed Systems Design

-
This is with apologies to Butler Lampson, who published the " Hints for computer system design " paper 40 years ago in SOSP'83. I don't claim to match that work of course. I just thought I could draft this post to organize my thinking about designing distributed systems and get feedback from others. I start with the same  disclaimer Lampson gave. These hints are not novel, not foolproof recipes, not laws of design, not precisely formulated, and not always appropriate. They are just hints.  They are context dependent, and some of them may be controversial. That being said, I have seen these hints successfully applied in distributed systems design throughout my 25 years in the field, starting from the theory of distributed systems (98-01), immersing into the practice of wireless sensor networks (01-11), and working on cloud computing systems both in the academia and industry ever since. These heuristic principles have been applied knowingly or unknowingly and has proven
Read more >>

Learning about distributed systems: where to start?

-
This is definitely not a "learn distributed systems in 21 days" post. I recommend a principled, from the foundations-up, studying of distributed systems, which will take a good three months in the first pass, and many more months to build competence after that. If you are practical and coding oriented you may not like my advice much. You may object saying, "Shouldn't I learn distributed systems with coding and hands on? Why can I not get started by deploying a Hadoop cluster, or studying the Raft code." I think that is the wrong way to go about learning distributed systems, because seeing similar code and programming language constructs will make you think this is familiar territory, and will give you a false sense of security. But, nothing can be further from the truth. Distributed systems need radically different software than centralized systems do.  --A. Tannenbaum This quotation is literally the first sentence in my distributed systems syllabus. Inst
Read more >>

Making database systems usable

-
Image
C. J. Date's Sigmod 1983 keynote, "Database Usability", was prescient. Usability is the most important thing to the customers. They care less about impressive benchmarks or clever algorithms, and more about whether they can operate and use a database efficiently to query, update, analyze, and persist their data with minimal headache. (BTW, does anyone have a link to the contents of this Sigmod'83 talk? There is no transcript around, except for this short abstract .) The paper we cover today is from Sigmod 2007. It takes on the database usability problem raised in that 1983 keynote head-on, and calls out that the king is still naked.  Let's give some context for the year 2007. Yes, XML format was still popular then. The use-case in the paper is XQuery. The paper does not contain any  reference to json. MongoDB would be released in 2009 with the document model; and that seems to be great timing for some of the usability pains mentioned in the paper! Web 2.0 was in
Read more >>

Looming Liability Machines (LLMs)

-
As part of our zoom reading group ( wow, 4.5 years old now ), we discussed a paper that uses LLMs for automatic root cause analysis (RCA) for cloud incidents. This was a pretty straightforward application of LLMs. The proposed system employs an LLM to match incoming incidents to incident handlers based on their alert types, predicts the incident's root cause category, and provides an explanatory narrative. The only customization is through prompt-engineering. Since this is a custom domain, I think a more principled and custom-designed  machine learning system would be more appropriate rather than adopting LLMs. Anyways, the use of LLMs for RCAs spooked me vicerally. I couldn't find the exact words during the paper discussion, but I can articulate this better now. Let me explain. RCA is serious business Root cause analysis (RCA) is the process of identifying the underlying causes of a problem/incident, rather than just addressing its symptoms. One RCA heuristic is asking 5 Why&#
Read more >>

Foundational distributed systems papers

-
I talked about the importance of reading foundational papers last week. To followup, here is my compilation of foundational papers in the distributed systems area. (I focused on the core distributed systems area, and did not cover networking, security, distributed ledgers, verification work etc. I even left out distributed transactions, I hope to cover them at a later date.)  I classified the papers by subject, and listed them in chronological order. I also listed expository papers and blog posts at the end of each section. Time and State in Distributed Systems Time, Clocks, and the Ordering of Events in a Distributed System. Leslie Lamport, Commn. of the ACM,  1978. Distributed Snapshots: Determining Global States of a Distributed System. K. Mani Chandy Leslie Lamport, ACM Transactions on Computer Systems, 1985. Virtual Time and Global States of Distributed Systems.  Mattern, F. 1988. Practical uses of synchronized clocks in distributed systems. B. Liskov, 1991. Expository papers
Read more >>

Advice to the young

-
Image
I notice I haven't written any advice posts recently. Here is a collection of my advice posts pre 2020. I've been feeling all this elderly wisdom pent up in me, ready to pour at any moment. So here it goes. Get ready to quench your thirst from my fount of wisdom. No man, think for yourself, only get what works for you. It is called foundations, not theory Foundations of computer science (or rather any field of study) are the most important topics you can learn. These lay down the frame of thinking/perspective for that area of study. Yet, I am saddened to hear these called as "theory", and labeled as "unpractical". This couldn't be farther from the truth. Take a look at how I recommend studying distributed systems . Don't you dare call this "theory" and "unpractical". This lays the bedrock that you build your practice on. Don't skimp on the foundations. Don't build your home on quicksand. Keep your hands dirty, your mind cl
Read more >>

Linearizability: A Correctness Condition for Concurrent Objects

-
Image
This paper is from Herlihy and Wing appeared in ACM Transactions on Programming Languages and Systems 1990. This is the canonical reference for the linearizability definition. I had not read this paper in detail before, so I thought it would be good to go to the source to see if there are additional delightful surprises in the original text. Hence, this post. I will dive into a technical analysis of the paper first, and then discuss some of my takes toward the end. I had written an accessible explanation of linearizability earlier; you may want to read that first. I will assume an understanding of linearizability to keep this review at reasonable length. Introduction I love how the old papers just barge in with the model, without bothered by pleasantries such as motivation of the problem. These are the first two sentences of the introduction. "A concurrent system consists of a collection of sequential processes that communicate through shared typed objects . This model encompass
Read more >>

Scalable OLTP in the Cloud: What’s the BIG DEAL?

-
Image
This paper is from Pat Helland, the apostate philosopher of database systems, overall a superb person, and a good friend of mine. The paper appeared this week at CIDR'24. (Check out the program for other interesting papers). The motivating question behind this work is: " What are the asymptotic limits to scale for cloud OLTP (OnLine Transaction Processing) systems? " Pat says that the CIDR 2023 paper "Is Scalable OLTP in the Cloud a Solved Problem?" prompted this question.  The answer to the question? Pat says that the answer lies in the joint responsibility of database and the application. If you know of Pat's work, which I have summarized several in this blog , you would know that Pat has been advocating along these lines before. But this paper provides a very crisp, specific, concrete answer. Read on for my summary of the paper. Disclaimer: This is a wisdom and technical information/detail packed 13-page paper, so I will try my best to summarize the sa
Read more >>

Understanding the Performance Implications of Storage-Disaggregated Databases

-
Image
Storage-compute disaggregation in databases has emerged as a pivotal architecture in cloud environments, as evidenced by Amazon ( Aurora ), Microsoft ( Socrates ), Google (AlloyDB), Alibaba ( PolarDB ), and Huawei (Taurus). This approach decouples compute from storage, allowing for independent and elastic scaling of compute and storage resources. It provides fault-tolerance at the storage level. You can then share the storage for other services, such as adding read-only replicas for the databases. You can even use the storage level for easier sharding of your database. Finally, you can also use this for exporting a changelog asynchronously to feed into peripheral cloud services, such as analytics. Disaggregated architecture was the topic of Sigmod 23 panel . I think this quote summarizes the industry's thinking on the topic. "Disaggregated architecture is here, and is not going anywhere. In a disaggregated architecture, storage is fungible, and computing scales independently.
Read more >>

Designing Data Intensive Applications (DDIA) Book

-
Image
We started reading this book as part of Alex Petrov's book club . We just got started, so you can join us, by joining the discord channel above. We meet Wednesday's 11am Eastern Time.  Previously we had read transaction processing book by Grey and Reuters. This page links to my summaries of that book. Chp 1. Reliable, Scalable, and Maintainable Applications I love the diagrams opening each chapter. Beautiful! The first chapter consists of warm up stuff. It talks about the definitions of reliabilty, scalability, and maintainability. It is still engaging, because  it is written with an educator and technical blogger voice, rather than a dry academic voice. This book came out on 2017. Martin is working on the new version. So if you have comments for things to focus on for the new version, it would be helpful to collect them in a document and email it to Martin. For example, I am curious about how the below paragraph from the Preface will get revised with 8 more years of hindsight:
Read more >>