360 likes | 409 Views
Divisible Load Scheduling. A Tutorial. Thomas Robertazzi University at Stony Brook. What is a Divisible Load?. A computational & networkable load that is arbitrarily partitionable (divisible) amongst processors and links. There are no precedence relations. Simple Application Example.
E N D
Divisible Load Scheduling A Tutorial Thomas Robertazzi University at Stony Brook
What is a Divisible Load? • A computational & networkable load that is arbitrarily partitionable (divisible) amongst processors and links. • There are no precedence relations.
Simple Application Example • Problem: Sum 1000 trillion numbers • Approach: Partition the numbers among 100 processors • But how?
Simple Application Example • To optimize solution time (maximize speedup) one needs to take into account heterogeneous link and processor speeds, computation and communication intensities, interconnection topology and scheduling policy. • Divisible Load Scheduling Theory Can Do This!
Applications (Generic) • Grid Computing/Meta-computing • Data Intensive Computing • Sensor Processing • Image Processing • Scientific/Engineering Computing • Financial Computing
Applications (Specific) • Pattern Searching • Database Computation • Matrix-Vector Computation • E&M Field Calculation (CAD) • Edge Detection
DLT Modeling Advantages • Linear and Deterministic Modeling • Tractable Recursive/Linear Equation Solution • Schematic Language • Equivalent Elements • Many Applications
Interconnection Topologies • Linear Daisy Chain • Bus • Single Level and Multilevel Trees • Mesh • Hypercube
Directions: Scalability 1 1 3 1 2 1 Simultaneous Distribution (Scalable) Sequential Distribution (Saturation) Cheng & Robertazzi Hung & Robertazzi
An Example • Model Specifications: • A star network( single level tree network), and multi-level tree. • Computation and transmission time is a linear function of the size of load. • Level to Level: Store and Forward Switching • Same Level: Concurrent Load Distribution.
Children without Front End: • After receiving the assigned data, each child proceeds to process the data.
Timing Diagram (single level tree) : • Children without Front End
m+1 unknows vs. m+1 Eqs. • Recursive equations: • Normalization equation:
The load distribution solution is similar to the solution of the state-dependent M/M/1 queuing system.
Similarities to Queueing Theory • Linear model and tractable solutions • Schematic Language • Equivalent Elelements • Infinite Size Networks
Tree Network • (Children without Front Ends)
Bandwidth of Fat Tree • Definition: The bandwidth of level j in a fat tree can be defined as pj-1z.
Directions: Sequencing and Installments • Daisy Chain Surprise • Efficiency Rule Ghose, Mani & Bharadwaj
Directions: Sequencing and Installments • Multi-installment for Sequential Distribution 6 4 5 3 1 2 Ghose, Mani & Bharadwaj
Directions: Sequencing and Installments Diminishing returns in using multi-installment distribution. Ghose, Mani & Bharadwaj
Directions: Sequencing and Installments Drozdowski
Directions: Time Varying Modeling Sohn & Robertazzi Can be solved with integral calculus.
Directions: Monetary Cost Optimization • Min CTotal = S ancnwnTcp N n=1 Bus Processors Optimal Sequential Distribution if: cn-1wn-1 less thancnwn for all n Sohn, Luryi & Robertazzi
Directions: Monetary Cost Optimization • 2 US Patents: Patent 5,889,989 (1999): Processor Cost Patent 6,370,560 (2001): Processor and Link Cost Enabling technology for an open e-commerce market in leased proprietary computing. Sohn, Charcranoon, Luryi & Robertazzi
Directions: Database Modeling Expected time to find multiple records in flat file database Ko & Robertazzi
Directions: Realism Finite Buffers (Bharadwaj) Job Granularity (Bharadwaj) Queueing Model Integration
Directions: Experimental Work Database Join (Drozdowski)
Directions: Future Research • Operating Systems: Incorporate divisible load scheduling into (distributed) operating systems • Measurement Process Modeling: Integrate measurement process modeling into divisible scheduling
Directions: Future Research • Pipelining (Dutot) Concept: Distribute load to further processors first for speedup improvement Improvement reported for daisy chains
Directions: Future Research • System Parameter Estimation (Ghose): Concept: Send small “probing” loads across links and to processors to estimate available effort Challenge: Rapid change in link & processor state
DLT has a Good Future • Many Applications including wireless sensor networks • Tractable (Modeling & Computation) • Rich Theoretical Basis
Thank you! Questions???