[Links of the day] 27/04/2015 : Containers at CERN, Nvidia Tesla and Usenix Hotcloud16

• Containers and Orchestration in the CERN Cloud : interesting talk looking at how CERN embrace containers on its cloud platform. And how they leverage Magnum which relies heavily on Heat for the orchestration part of the container clusters - called bays. I would be curious to find out how they scaled Heat and pitfall with this approach. 
• NVIDIA Tesla P100 : Nvidia whitepaper on the next generation accelerator . Impressive spec, however I always feel that there is still significant room for making it easier to consume those type of ressource.
• HotCloud16  : Usenix program is out, some interesting papers especially these two : 
• Cloud Spot Markets are Not Sustainable: The Case for Transient Guarantees
• Mlcached: Multi-level DRAM-NAND Key-value Cache

Links of the day 23/10/2015: #Openstack, #KVM, #Docker at scale at #CERN

• Openstack at scale at CERN : 5k hypoerisor, 15k VM, up to 1 Millions vm, the scale of the CERN openstack deployment is really what is driving the effort to push openstack beyond its limits. 
• HEPIX fall 2015 : Hepix is a really interesting workshop as it provide real use case report and experience of large scale IT infrastructure and as time evolved pretty much HPC cloud centric solution. 
• Performance  comparison of   KVM  and  Docker instances  in  OpenStack : well no surprise here, docker faster all the way

Links of the day 03/09/2015 : vNVDIMM, user-space tcp/ip stack, Cern openday

• Event-driven, user-space and highly-scalable TCP/IP stack : framework got a 2.6× performance improvement, when compared to the same application layer running on the new reusable TCP sockets introduced in Linux 3.9. The web-server was able to deliver static HTML pages at a rate of 12 Gbps on a Tilera TILE-Gx36 36-core processor. Userspace vs kernel space ..
• Cern Openlab Day : research and industrial teams from CERN openlab present their projects
• vNVDIMM : NVDIMM driver has been merged into upstream Linux Kernel and there is tries to enable it in virtualization field. Virtual machine with persistent memory.

Links of the day 02/09/2015 : Seagate kintetic @CERN , KVM HA and Intel Omnipath perf

• KVM HA : Update on the COLO VM replication project. What's interesting is that they only look at the network side of the external state to confirm the coherence of the system. This limits what the VM can do and access but on the other hand simplify the overall setup.
• Kinetic @ CERN : In a previous blog post I mentioned how this type of approach is a game changer for cloud storage solution. Here you can already see some numbers : ~1/3 of power , ~1/2 of cost, No more storage server ( simplified maintenance ) [video]
• Omnipath :  the number are really interesting: (a) a 4.6x improvement in small message throughput over the previous generation InfiniBand fabric technology, (b) a 70ns decrease in switch latency  (c) a single ASIC that can deliver 50 GB/s of dual-channel bidirectional bandwidth, or 12.5GB/s single channel unidirectional bandwidth. In short, Omni-Path delivers all this and more as an integrated component inside the processor package or via other form factors like PCIe cards or custom mezzanine cards.

Links of the day 03 - 04 - 2015

Today's links 03/04/2015: NVM trends and loads of conferences - Hepix, CernVm, Lambda Days

• Emerging Non Volatile Memory (NVM) Technology & Market Trends: 2014 Report by Yole Developpement
• HEPiX workshop : lots of very good talk in there.
• CernVM workshop : ecosystem providing all the building blocks for a virtualized analysis environment for High Energy Physics experiments. 
• Lambda 2015 : Functional Programming Conference (26-27 FEBRUARY 2015)

Links of the day 28 - 10 - 2014

Today's links 28/10/2014: cloud trading, CERN, snapshot isolation, cloud infrastructure, RDMA to remote HW/device

• Clock-SI : [paper summary] Snapshot Isolation for Partitioned Data Stores Using Loosely Synchronized Clocks
• Cloud trading : 6 fusion blog post on their approach on trading cloud resources as commodities. 
• CERN Cloud Infrastructure Report : a lot openstack, looks like they are trying to double the capacity every year. They are really at the bleeding edge of openstack production and stressing what can be done with it. 
• Peer-Direct support : allows RDMA operations to directly target memory in external hardware devices, such as GPU cards, SSD based storage, dedicated ASIC accelerators, etc.