Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userA Generic Checkpoint-Restart Mechanism for Virtual Machines
114
0
0.0
(
0
)
Added by
Rohan Garg
Publication date
2012
fields
Informatics Engineering
and research's language is
English
Ask ChatGPT about the research
No Arabic abstract
It is common today to deploy complex software inside a virtual machine (VM). Snapshots provide rapid deployment, migration between hosts, dependability (fault tolerance), and security (insulating a guest VM from the host). Yet, for each virtual machine, the code for snapshots is laboriously developed on a per-VM basis. This work demonstrates a generic checkpoint-restart mechanism for virtual machines. The mechanism is based on a plugin on top of an unmodified user-space checkpoint-restart package, DMTCP. Checkpoint-restart is demonstrated for three virtual machines: Lguest, user-space QEMU, and KVM/QEMU. The plugins for Lguest and KVM/QEMU require just 200 lines of code. The Lguest kernel driver API is augmented by 40 lines of code. DMTCP checkpoints user-space QEMU without any new code. KVM/QEMU, user-space QEMU, and DMTCP need no modification. The design benefits from other DMTCP features and plugins. Experiments demonstrate checkpoint and restart in 0.2 seconds using forked checkpointing, mmap-based fast-restart, and incremental Btrfs-based snapshots.
rate research
Read More
InfiniBand is widely used for low-latency, high-throughput cluster computing. Saving the state of the InfiniBand network as part of distributed checkpointing has been a long-standing challenge for researchers. Because of a lack of a solution, typical MPI implementations have included custom checkpoint-restart services that tear down the network, checkpoint each node as if the node were a standalone computer, and then re-connect the network again. We present the first example of transparent, system-initiated checkpoint-restart that directly supports InfiniBand. The new approach is independent of any particular Linux kernel, thus simplifying the current practice of using a kernel-based module, such as BLCR. This direct approach results in checkpoints that are found to be faster than with the use of a checkpoint-restart service. The generality of this approach is shown not only by checkpointing an MPI computation, but also a native UPC computation (Berkeley Unified Parallel C), which does not use MPI. Scalability is shown by checkpointing 2,048 MPI processes across 128 nodes (with 16 cores per node). In addition, a cost-effective debugging approach is also enabled, in which a checkpoint image from an InfiniBand-based production cluster is copied to a local Ethernet-based cluster, where it can be restarted and an interactive debugger can be attached to it. This work is based on a plugin that extends the DMTCP (Distributed MultiThreaded CheckPointing) checkpoint-restart package.
Providing fault-tolerance for long-running GPU-intensive jobs requires application-specific solutions, and often involves saving the state of complex data structures spread among many graphics libraries. This work describes a mechanism for transparent GPU-independent checkpoint-restart of 3D graphics. The approach is based on a record-prune-replay paradigm: all OpenGL calls relevant to the graphics driver state are recorded; calls not relevant to the internal driver state as of the last graphics frame prior to checkpoint are discarded; and the remaining calls are replayed on restart. A previous approach for OpenGL 1.5, based on a shadow device driver, required more than 78,000 lines of OpenGL-specific code. In contrast, the new approach, based on record-prune-replay, is used to implement the same case in just 4,500 lines of code. The speed of this approach varies between 80 per cent and nearly 100 per cent of the speed of the native hardware acceleration for OpenGL 1.5, as measured when running the ioquake3 game under Linux. This approach has also been extended to demonstrate checkpointing of OpenGL 3.0 for the first time, with a demonstration for PyMol, for molecular visualization.
Fault tolerance for the upcoming exascale generation has long been an area of active research. One of the components of a fault tolerance strategy is checkpointing. Petascale-level checkpointing is demonstrated through a new mechanism for virtualization of the InfiniBand UD (unreliable datagram) mode, and for updating the remote address on each UD-based send, due to lack of a fixed peer. Note that InfiniBand UD is required to support modern MPI implementations. An extrapolation from the current results to future SSD-based storage systems provides evidence that the current approach will remain practical in the exascale generation. This transparent checkpointing approach is evaluated using a framework of the DMTCP checkpointing package. Results are shown for HPCG (linear algebra), NAMD (molecular dynamics), and the NAS NPB benchmarks. In tests up to 32,752 MPI processes on 32,752 CPU cores, checkpointing of a computation with a 38 TB memory footprint in 11 minutes is demonstrated. Runtime overhead is reduced to less than 1%. The approach is also evaluated across three widely used MPI implementations.
Unified Virtual Memory (UVM) was recently introduced on recent NVIDIA GPUs. Through software and hardware support, UVM provides a coherent shared memory across the entire heterogeneous node, migrating data as appropriate. The older CUDA programming style is akin to older large-memory UNIX applications which used to directly load and unload memory segments. Newer CUDA programs have started taking advantage of UVM for the same reasons of superior programmability that UNIX applications long ago switched to assuming the presence of virtual memory. Therefore, checkpointing of UVM will become increasingly important, especially as NVIDIA CUDA continues to gain wider popularity: 87 of the top 500 supercomputers in the latest listings are GPU-accelerated, with a current trend of ten additional GPU-based supercomputers each year. A new scalable checkpointing mechanism, CRUM (Checkpoint-Restart for Unified Memory), is demonstrated for hybrid CUDA/MPI computations across multiple computer nodes. CRUM supports a fast, forked checkpointing, which mostly overlaps the CUDA computation with storage of the checkpoint image in stable storage. The runtime overhead of using CRUM is 6% on average, and the time for forked checkpointing is seen to be a factor of up to 40 times less than traditional, synchronous checkpointing.
The share of the top 500 supercomputers with NVIDIA GPUs is now over 25% and continues to grow. While fault tolerance is a critical issue for supercomputing, there does not currently exist an efficient, scalable solution for CUDA applications on NVIDIA GPUs. CRAC (Checkpoint-Restart Architecture for CUDA) is new checkpoint-restart solution for fault tolerance that supports the full range of CUDA applications. CRAC combines: low runtime overhead (approximately 1% or less); fast checkpoint-restart; support for scalable CUDA streams (for efficient usage of all of the thousands of GPU cores); and support for the full features of Unified Virtual Memory (eliminating the programmers burden of migrating memory between device and host). CRAC achieves its flexible architecture by segregating application code (checkpointed) and its external GPU communication via non-reentrant CUDA libraries (not checkpointed) within a single processs memory. This eliminates the high overhead of inter-process communication in earlier approaches, and has fewer limitations.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
