Salesforce Com, Inc.

Andrey Yegorov

Venkateswararao Jujjuri

Rithin Kumar Shetty

In some embodiments, a server node may transmit metrics to a client piggy-backed on (appended to) responses to regular read/write requests. Other embodiments may be disclosed and/or claimed.

Dissemination of node metrics in server clusters

Among heterogeneous nodes in a cluster, one node's capabilities can vary widely from that of others in the same cluster, for example, in a cluster of storage servers. As each of the clients/writers to the cluster may contact different server nodes in the cluster over time, with each contact the client/writer may receive server metric information, including both the contacted node's current metrics as well as metrics for any other nodes of which the contacted node is aware. The client/writer may provide updated information to the contacted node regarding the metrics for other nodes of which the client/writer is aware or has contacted. The exchange of metric information may be piggybacked upon read/write requests to each node to conserve bandwidth.

Opportunistic gossip-type dissemination of node metrics in server clusters

Distributed Systems are becoming very large and are being served by 10s of 1000s of individual machines. Among heterogeneous nodes, one node's capabilities can vary widely from that of others in the same cluster, for example, in a cluster of storage servers. Each of the clients/writers may retrieve, periodically, from all the servers, important metrics such as free disk space capacity, average disk IO utilization over the last N minutes, etc. We call these client to nodes (servers) messages as 'periodic metrics query.' Using received metrics, clients dynamically assign weights to each of the servers as an indication of their current capabilities. Server metrics and or weights are distributed across the clients, and each client may make independent placement decisions based on their view of the cluster. Because clients are using substantially the same information for making decisions on placement, the result is improved balance across the servers.

Growing community of inventors

San Francisco, CA, United States of America

Andrey Yegorov