The Future of Virtualization: The Virtualization of the Future Larry Rudolph, VMware 200 VMware Inc. All rights reserved
Caveats The opinions expressed in this talk are Entirely my own Do not come with any guarantees Subject to interpretation Based on years of experience having little to do with virtualization (I joined VMWare a year ago but I am still an academic ) 2 2
Outline -- Disembodied Computation Beyond the OS More migration Beyond PC s and Servers bigger, smaller Beyond Computers Don t let the cat out of the bag No cat, no bag. No, really 3 3
Process Priority OS Scheduler In a time-sharing want shortest jobs first cpu intensive jobs run in background Now think out of the box I/O Bound Compute Bound Time Quantum Length 4 4
Location Scheduler In a rich environment Jobs should move near their most critical resources I/O devices Storage Compute Devices Resource Allocation You can t always get what you want, but if you try sometime, you get what you need. 5 5
Load Balancing Socialism, Capitalism, or Environmentalism? Marginal Utility Curve per job How does job s performance vary with its location Location Marginal Utility Curve per Load increase number of jobs at a location may slow all Solve by initial guess, then iterate 6 6
Virtualization makes this possible Start job on local PC: If it needs more compute power, migrate it to server migrate it to supercomputer If it is always accessing a remote file system, migrate it to better SAN access migrate it to data center / cloud If it needs user interaction, Assumptions: * Virtualization nearly everywhere * Authorization to use * etc. migrate to display If multiple high needs, split or thrash 7 7
Review: Benefits of Virtalization Isolation no connection between virtual machines Disembodiment The app & OS disconnected from HW CPU, MMU, Devices from each other Resource Sharing Time sharing done right 8 8
High Performance Computing HPC want every cycle of performance for their apps suspicious of virtual memory; VMs take too much overhead HPC applications optimally balance computation with communication virtualization that hides or modifies the ratio --> LPC (Low PC) HPC applications often require low-latency communication barrier synchronizations, collective ops such as reduce data-centers care about bandwidth and throughput, not latency e.g. infiniband not supported under current virtualization offerings HPC programmers use special purpose (expensive) supercomputers 9 9 9
HPC and the Cloud Despite negatives, there is good reasons to support HPC High-end HPC mostly government funded; just enough machines to meet current needs Desire for elastic resource more computational scientists in academic & industry Massive data centers will have 00,000 s of cores and are efficiently managed Co-locate small (windows-based) computations that feed into HPC task e.g. analyze biological slides on PC, then global analysis Fast decision making based on models -- e.g. finacial 0 0
Big Computation Today s Supercomputers have 00,000 s cores, hi bandwidth, lo latency Thrilling to use, but infrequent Supercomputing for the masses languages, packages, tools all in place Need seemless scaling from S -> M-> L Need to debug/refine only small part Create VMs, one per core, under-full simulation record neighbor communication (small fraction of time) debugging: full sim. island nodes, synthetic execute rest when effects changes neigh, change them to real full simulation.
Petascale Ocean Model Numbers Div ide simulation domain into subregions (white lines on picture). Each subregion cov ers n (256 in this exercise) processors. Sav e messages at subregion boundaries To replay one y ear of simulation f or region marked ŅÓabov e. Would require ~300GB of stored data (assuming no user directed optimization) and ~256 processors. This would enable an exact rerun of region ŅÓf or v isualization or analy sis of ~7TB of data. For the whole simulation archiv e requirement is ~35TB, to allow any region to be replay ed. 2
Little computations 3
Larry s Mission Open Please Phone Stakeholders: Handset Manufacturer Carrier ISV Where does the owner fit it? When one buys a phone, one should be able to install any OS on it, e.g. root password Clouds -- Vertically integrated, implementation details hidden. Academic research relegated to toy systems? 4
Virtual-physical boundary Remember Do-over s? I wish life was like that E.g. sent email but wish you didn t? When do actions commit? when they are observable in the physical world someone reads their email, output to printer, launch missle checkpoint & rollback but across a system VM s on distinct machines may cooperate to roll-back 5
Virtual Machines meet Virtual Reality 6
Mixed Reality 2nd life on handheld or handheld in 2nd life 7 7
Blurring the boarders Seamless move between virtual and real world Human world switch is expensive Virtual world can keep track of all activities for later reference eliminates any need to sync between all different devices We can use whatever (computer-based) device is convenient The virtualization of the future 8 Confidential
9 9