A Distributed Systems Query Agenda For Cloud Computing

Distributed systems (a.k.a. distributed algorithms) is an one-time land of almost forty years old. It gave us impossibility proofs on the theory side, together with likewise algorithms similar Paxos, logical/vector clocks, 2/3-phase commit, leader election, dining philosophers, graph coloring, spanning tree structure which are adopted inwards do widely. Cloud computing is a relatively novel land inwards contrast. It provides novel opportunities every bit good every bit novel challenges for the distributed systems/algorithms area. Below I briefly beak over around of these opportunities together with challenges.

Opportunities

Cloud computing provides abundance. Nodes are replaceable, fifty-fifty hot swappable. You tin terminate dedicate several nodes for running customized back upward services, such every bit monitoring, logging, storage/recovery service. These opportunities are probable to bring ship upon on how fault-tolerance is considered inwards distributed systems/algorithms work.

Programmatic interfaces together with service-oriented architecture are likewise hallmarks of cloud computing domain. Similarly, virtualization/containerization facilitates many things, including moving the computation to the data. However, it is unclear how these technologies tin terminate travel employed to supply substantial benefits for distributed algorithms which operate at a to a greater extent than abstract plane.

Challenges

Coordination introduces synchronization, which introduces potential for cascading shutdowns together with halts.  Especially, inwards a large-scale organisation of systems, whatever coordination introduced may Pb to spooking/latent/self-organized synchronization. This holler for is explained nicely inwards "Towards A Cloud Computing Research Agenda". Thus the challenge is to avoid coordination every bit much every bit possible, together with all the same build useful together with "consistent" systems.

Another curse of the extreme scale inwards cloud computing is the large fan-in/fan-out of components. These challenges are explained nicely inwards due to the CAP theorem.

The electrical current state of distributed systems query on cloud computing

This is non an exhaustive list. From what I tin terminate recall now, hither is a unsmooth categorization of electrical current query inwards distributed systems on the cloud computing domain.

• Consistency inwards geo replicated systems, e.g., COPS, Red Blue Consistency, Use of clocks for consistency, HLC
• Paxos variants, e.g., Mencius, Tango, Chain replication
• Efficient transaction primitives, e.g., Sinfonia, Salt, Calvin, Spanner, Partitioned replicated cloud databases
• Data pump networking, e.g., Onix, VL2, and many recent work.
• CAP boundaries, e.g., Consistency, availability, together with convergence
• CRDTs, e.g.,  Conflict-free replicated information types, Eventual consistency linguistic communication frameworks, Coordination Avoidance inwards Database Systems
• New programming abstractions for large information together with large graph processing, e.g., Naiad GraphX, Pregel

What is next?

If solely I knew... I tin terminate solely speculate, I bring around ideas, together with that volition bring to await for around other post.

In the meanwhile, I tin terminate holler for to these 2 articles which talked almost what would travel good/worthy query areas inwards cloud computing.
How tin terminate academics create query on cloud computing?
Academic cloud: pitfalls, opportunities

UPDATE: Part 2 of this postal service (i.e., novel directions) is available here. 

Share this post