320 likes | 460 Views
Towards the Future of IT. does the path lead through the cloud?. Lutz Schubert, Institute of Information Resource Management, University of Ulm High Performance Computing Center Stuttgart. The Future of Information Technology. A Little History Lesson.
E N D
Towards the Future of IT does the path lead through the cloud? Lutz Schubert, Institute of Information Resource Management, University of Ulm High Performance Computing Center Stuttgart
The Future of Information Technology A Little History Lesson “Where we cannot invent, we may at least improve.” Charles Caleb Colton
45 40 35 30 25 20 15 10 5 0 Uptake of the Internet number of users Introducing Load Balancing (milltions) number of hosts
400 45 350 40 35 300 30 250 25 200 20 150 15 100 10 50 5 0 0 Uptake of the Internet number of users (milltions) number of hosts
400 350 300 250 200 150 100 50 0 Uptake of the Internet number of users Uptake of Outsourcing (milltions) number of hosts load balancing
400 1000 350 300 800 250 600 200 150 400 100 200 50 0 0 Uptake of the Internet number of users (milltions) number of hosts
1000 800 600 400 200 0 Uptake of the Internet number of users The Grid & Clouds (milltions) outsourcing number of hosts load balancing
2000 1000 800 1500 600 1000 400 500 200 0 0 Uptake of the Internet number of users (milltions) number of hosts
2000 1500 1000 500 0 Uptake of the Internet number of users ??? (milltions) grids & clouds number of hosts outsourcing load balancing
2000 1500 1000 500 0 Uptake of the Internet number of users (milltions) heterogeneity grids & clouds number of hosts multicores smart phones outsourcing GPGPUs load balancing
At the same time… • The value of internet services keeps growing… Source: http://www.atkearney.com/paper/-/asset_publisher/dVxv4Hz2h8bS/content/internet-value-chain-economics/10192
Strong Increase • Mobility • Heterogeneity • Amount of Data • #Users The Trend Continues • More users (only 25% online now) • More devices (also the washing machine goes online) • More mobility, within and across borders • More global usage • Less Increase • Performance • Communication • Storage • (User) Expectations • Availability • Speed • Cost • Ease of use
A Quick Report from The Cloud Expert Group The Problems http://cordis.europa.eu/fp7/ict/ssai/docs/ future-cc-2may-finalreport-experts.pdf “All change is not growth; as all movement is not forward.” Ellen Glasgow
#1 Availability ≠ Speed 1 kB ~2ms* 1 GB ~10 mins* 1 TB ~1 week* 1 kB ~0.1 s* 1 GB ~1 day* 1 TB ~2 years* *purely theoretical value not allowing for average network load, different providers etc.
#2 Dynamicity vs. Security? Help, mydatais in China?
#3 The Whole Application Moves 1 kB ~2ms* 1 GB ~10 mins* 1 TB ~1 week* 1 kB ~0.1 s* 1 GB ~1 day* 1 TB ~2 years* *purely theoretical value not allowing for average network load, different providers etc.
#3 each part exhibits different needs… 1 kB ~2ms* 1 GB ~10 mins* 1 TB ~1 week* frequently accesseddata consistency? 1 kB ~0.1 s* 1 GB ~1 day* 1 TB ~2 years* rarelyaccesseddata
#3 … use them • Separate user input • Shared state / data • Highly volatile data • Compute intensive • Points of little dependencies • Google specific API call Bottleneck Can be replicated Needs specific resources
#3 … use them • Reduced resource cost • Not all application parts need to scale out fully (less resources) • Faster adaptation time (better dynamicity) • The individual requirements are lower (simpler resources) • Higher customer satisfaction • Better fulfillment of availability etc. • Trust & security through hybrid (public-private) clouds
double heatmap[10]; double heatmap_tmp[10]; #pragma func main() = heatspread( initmap( heatmap ) ) void main() { initmap(*heatmap); for (int iter=0; iter<100; iter++) // 100 iterations { heatspread(*heatmap); memcpy(heatmap, heatmap_tmp, 10) } Specifies output dependencies } } } Specifies input data dependencies #pragma func heatspread(heatmap) = zipWith (+) (heatcell( neighbours(heatmap) )) (heatmap) void heatspread (double** heatmap) { Specifies for (int x=0; x<10; y++) workflow { double dphi = heatcell( neighbours(heatmap, x) ); heatmap_tmp[x] = heatmap[x] + dphi; } } ... #4 Programmability
Conclusions http://cordis.europa.eu/fp7/ict/ssai/docs/cloud-expert-group/ roadmap-dec2012-vfinal.pdf “My interest is in the future because I am going to spend the rest of my life there.” Charles F. Kettering
Cloud Development Roadmap demanding services on the fly federated & merged clouds bring your own device rich composed services collaboration eEngineering brokering personal clouds sensor clouds mobile clouds eGovernment carry your environment auto-composition eHealth eScience exascale new resources BYOD personalisation globalisation service-centric huge data new IT mid term long term short term
The Future is Cloudy? • Future environments are • Even more heterogeneous on all levels (specialisation) • Even more widely dispersed (global mobility) • Large scale with little performance per unit (multicore) • Demanding more complex usage scenarios • Clouds are a first reaction to this paradigm change • They are not the FINAL solution • But they are a necessary transition into the future
THANKS FOR LISTENING Lutz Schubert Institute of Resource Management, University of Ulm lutz.schubert@uni-ulm.de