By a News Reporter-Staff News Editor at Life Science Weekly -- Symantec Corporation (Mountain View, CA) has been issued patent number 10015266, according to news reporting originating out of Alexandria, Virginia, by NewsRx editors (see also Symantec Corporation).
The patent's inventors are Shavell, Michael (Merrimack, NH); Newstadt, Keith (West Newton, MA).
This patent was filed on September 22, 2014 and was published online on July 3, 2018.
From the background information supplied by the inventors, news correspondents obtained the following quote: "A contemporary cloud based datacenter can make applications, services and data available to very large numbers of client computers simultaneously. For example, a given node in the datacenter can have millions of concurrent network connections. Clusters are groups of computers that can be deployed in these can other contexts to provide high availability. Clusters use groups of redundant computing resources in order to provide continued service when individual system components fail. Clusters eliminate single points of failure by providing multiple servers, multiple network connections, redundant data storage, etc. Clustering systems are often combined with storage management products that provide additional useful features, such as journaling file systems, logical volume management, etc.
"It is also the case that the computing resources utilized by a datacenter change over time as a result of maintenance, decommission and upgrades of existing resources, installation of new resources, etc. In addition, the demand for resources also varies over time based on levels of client and internal activity, which in turn drives needed resource capacity. As the available resources change and levels of desired capacity increase and decrease, load balancing is used to distribute workloads across computing resources such as nodes of the cluster and storage devices. Load balancing attempts to optimize resource use based on factors such avoiding maintenance of overhead capacity, speed, maximization of throughput, etc.
"When a single node has a large number of connections (e.g., millions), all of these connections are severed if the node is taken down (for example, because of detected instability, for maintenance or replacement, to address a decrease in demand, etc.). When this occurs, the large number of client computers with severed connections to the node receive connection failure errors, causing them to reissue their requests. This results in the datacenter getting bombarded with requests, and creates a major overhead often resulting in over-allocation of resources to address the rapid re-connect scenario. It also creates a latency for the clients as they try to get responses to their requests, which leads to a degraded experience for the end users of the client computers. This problem affects all network connection types, but is particularly acute for types of connections that are kept open between computers for a longer period of time (known as 'long lived connections') such as long poll, Hypertext Transfer Protocol ('HTTP') streaming, and HTTP pipelining connections.
"It would be desirable to address these issues."
Supplementing the background information on this patent, NewsRx reporters also obtained the inventors' summary information for this patent: "Large numbers (e.g., a million or more) connections are migrated between a source computer and one or more destination computers in a computer network, such as a cluster or datacenter. The source and destination computers can be in the form of servers in a datacenter which provides services and the like to endpoint computers (e.g., remotely located clients). The connections between a source computer and multiple endpoints that are migrated to one or more destination computers can be in the form of long lived connections, such as long poll, HTTP streaming and/or HTTP pipelining connections. The connections are maintained during the migration without stopping traffic between the network (e.g., the source and destination computers) and the endpoint computers.
"Multiple connections between multiple endpoint computers and a specific source computer in the network are migrated from the source computer to one or more destination computers in the network. The migration can be initiated, for example, in response to the source computer receiving a corresponding control signal from the network load balancer. In another embodiment, the migration is initiated automatically by the source computer, for example in response to the results of self-diagnostics, or according to a pre-determined schedule. In another embodiment, the migration is initiated in response to an indication input by a human (e.g., entered by an administrator operating a GUI).
"Connection state information concerning each request from each one of the multiple endpoint computers being processed on each one of the multiple connections is transferred from the specific source computer to the destination computer, in response to the initiating of the migration. The source computer continues to respond to requests received on a specific connection while transferring corresponding connection state information to the specific destination computer. Once connection state information for a specific connection has been transferred to the destination computer, the destination computer is notified that it is now responsible for the specific connection, and the load balancer is notified to switch the specific connection from the source computer to the destination computer. In one embodiment, these notifications are performed separately for each specific connection, subsequently to transferring the corresponding connection state information to the destination computer. In other embodiments, the notifications are made in batch mode for a plurality of specific connections, after the corresponding connection state information for each of the connections has been transferred.
"The traffic to and from the multiple endpoint computers on the multiple connections is shifted from the source computer to the destination computer, while the multiple connections are kept active during the shifting. In other words, the traffic between the multiple endpoint computers and the network (e.g., datacenter) does not stop during the migration of the multiple connections from the source computer to the destination computer. While shifting traffic on a specific connection from the source to the destination computer, two open versions of the specific connection can be maintained simultaneously, one to the source computer and the other to the destination computer. Traffic on the specific connection can be linked to both the source and destination computer during the shifting process. The load balancer can monitor for traffic on the connection during this time, and route the monitored traffic to the source computer until the shift has been completed. The completion of the shift to the destination computer can be performed while the specific connection is in a quiet wait state. After having completed the shifting of a specific connection from the source computer to the destination computer, the connection can be closed on the source computer, and the traffic routed to the destination computer on its open connection.
"In some embodiments, the connections are migrated from a single source computer to multiple destination computers. The destination computer(s) to which to migrate connections can be determined based on factors such as availability, available network bandwidth, available processing capacity, available storage and proximity to endpoint computers. In some embodiments, connections are simultaneously migrated from multiple source computers to one or more destination computers.
"The features and advantages described in this summary and in the following detailed description are not all-inclusive, and particularly, many additional features and advantages will be apparent to one of ordinary skill in the relevant art in view of the drawings, specification, and claims hereof. Moreover, it should be noted that the language used in the specification has been principally selected for readability and instructional purposes, and may not have been selected to delineate or circumscribe the inventive subject matter, resort to the claims being necessary to determine such inventive subject matter."
For the URL and additional information on this patent, see: Shavell, Michael; Newstadt, Keith. Live Migration of Massive Numbers of Network Connections, Including Long Lived Connections. U.S. Patent Number 10015266, filed September 22, 2014, and published online on July 3, 2018. Patent URL: http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO1&Sect2=HITOFF&d=PALL&p=1&u=%2Fnetahtml%2FPTO%2Fsrchnum.htm&r=1&f=G&l=50&s1=10015266.PN.&OS=PN/10015266RS=PN/10015266
Keywords for this news article include: Business, Computers, Software Companies, Network Connections, Symantec Corporation, Technology Companies, Internet Software and Service Companies.
Our reports deliver fact-based news of research and discoveries from around the world. Copyright 2018, NewsRx LLC