Distributed operating system functions for nodes in rack

    公开(公告)号:GB2560141B

    公开(公告)日:2019-02-13

    申请号:GB201810591

    申请日:2016-12-19

    Applicant: IBM

    Abstract: A computer-implemented method includes managing function calls between a plurality of nodes and a super node of a rack system having a distributed operating system (OS). The OS includes a plurality of functions divided into first class and a second class, and each of the plurality of nodes excludes functions in the second class. Managing the function calls includes detecting a call to a first function on a first node of the plurality of nodes. It is determined that the first function belongs to the second class of functions and is not available on the first node. The call to the first function is routed to the super node, responsive to determining that the first function belongs to the second class, where the super node includes code for the functions in the second class.

    Checkpointing module and method for storing checkpoints

    公开(公告)号:GB2533342A

    公开(公告)日:2016-06-22

    申请号:GB201422436

    申请日:2014-12-17

    Applicant: IBM

    Abstract: A checkpoint module receives check-points from computer modules through interface 11 and comprises a memory management entity 13 assigning storage locations within non-volatile random access memories (RAM) 12 and volatile RAM 16. Memory controller 14 stores the received checkpoints to the assigned locations unifying different types of RAM within a uniform address space. Other modules comprise source code entity 15 for lossless elimination of redundant information via data differencing, differential compression, or data compression. Data differencing is based on a reference checkpoint, while data compression is performed in the absence of the reference checkpoint. Power management 18 disables source coding of a subset of checkpoints upon power outage. External storage devices with block access storage, e.g. magnetic tapes, rotating and solid-state disks, are used in addition to memories 12, 16 via external storage interface 19. The non-volatile memory can be a phase change memory (PCM) cell or a memristor cell.

    Distributed operating system functions for nodes in rack

    公开(公告)号:GB2560141A

    公开(公告)日:2018-08-29

    申请号:GB201810591

    申请日:2016-12-19

    Applicant: IBM

    Abstract: A computer-implemented method includes managing function calls between a plurality of nodes (120) and a super node (130) of a rack system (100) having a distributed operating system (OS). The OS includes a plurality of functions divided into first class (160) and a second class (165), and each of the plurality of nodes (120) excludes functions in the second class (165). Managing the function calls includes detecting a call to a first function on a first node of the plurality of nodes (120). It is determined that the first function belongs to the second class (165) of functions and is not available on the first node. The call to the first function is routed to the super node (130), responsive to determining that the first function belongs to the second class (165), where the super node (130) includes code for the functions in the second class (165).

Patent Agency Ranking