IP Library Granted Patent US 8,694,575
Granted Patent B2
US 8,694,575 · App. 13/675,419 · Granted Apr 8, 2014

Data-intensive computer architecture

Inventors: Edward Givelberg (Baltimore, MD); Sandor Szalay (Baltimore, MD)
Assignee: The Johns Hopkins University
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 8,694,575
App. No.
13/675,419
Granted
Apr 8, 2014
Kind
B2
Abstract

A data-intensive computer includes a processing sub-system formed by a plurality of processing servers, a database sub-system formed by a plurality of database servers configured to form a collective database, and a unifying operating system environment. The unifying operating system environment extends across the processing sub-system and the database sub-system to coordinate operation of the plurality of processing servers and the plurality of database servers to act as components in a single data-intensive computer and presents the database sub-system to an application running in the data-intensive computer as a layer in a memory hierarchy of the data-intensive computer.

Claims (30)

1. A data-intensive computer comprising:

a processing sub-system formed by a plurality of processing servers;

a database sub-system formed by a plurality of database servers configured to form a collective database;

a unifying operating system environment extending across the processing sub-system and the database sub-system to coordinate operation of the plurality of processing servers and the plurality of database servers to act as components in a single data-intensive computer;

wherein the unifying operating system environment is further configured to present the database sub-system to an application running in the data-intensive computer as a layer in a memory hierarchy of the data-intensive computer;

wherein the unifying operating system environment is further configured to implement a memory system for the data-intensive computer to deliver pages between the processing sub-system and the database sub-system such that the plurality of database servers provide 0(log N) access to individual random pages of data stored within the collective database storing N pages of data; and

wherein the unifying operating system environment presents a system management layer, a data object layer, a database access layer, and a data transport layer to coordinate operation of the processing sub-system and the database sub-system.

2. The data-intensive computer of claim 1 wherein the unifying operating system environment is configured to monitor data requirements of the application and facilitate access to the database sub-system based on the data requirements of the application.

3. The data-intensive computer of claim 1 further comprising a plurality of operating system servers acting as a distributed scheduler for the plurality of database servers to dynamically allocate and reduce connections to selected ones of the plurality of database servers in the database sub-system based on data requirements of the application.

4. The data-intensive computer of claim 3 wherein the operating system sub-system is configured to facilitate transparent and scalable input/output (I/O) operations between the plurality of processing node servers and the plurality of database servers.

5. The data-intensive computer of claim 1 wherein the unifying operating system environment is configured to logically represent a specific data object stored in the database sub-system as a single data object when the specific data object is physically distributed across the plurality of database servers of the database sub-system.

6. The data-intensive computer of claim 5 wherein the unifying operating system environment is configured to generate a physical mapping of the specific data object physically distributed across the plurality of database servers of the database sub-system and utilize the physical mapping to present the specific data object as the single data object to the application.

7. The data-intensive computer of claim 1 wherein the data transport layer is configured to manage connections by client computers accessing the data-intensive computer over a network external to the data-intensive computer using at least two independent implementations of the data transport layer, wherein a first implementation of the data transport layer uses UDP-based Data Transfer (UDT) sockets.

8. The data-intensive computer of claim 7 wherein a second implementation of the data transport layer uses message passing interface 2 (MPI-2) protocols.

9. The data-intensive computer of claim 7 wherein the unifying operating system environment presents a server software package and a client software package and wherein the sever software package and the client software package are implemented using at least one of the UDT sockets and the MPI protocols.

10. The data-intensive computer of claim 1 wherein the database access layer facilitates execution of queries and access requests implicating the database sub-system.

11. The data-intensive computer of claim 1 wherein the data object layer contains a description of user-defined programming objects stored in the database sub-system.

12. The data-intensive computer of claim 11 wherein the description of user-defined programming objects includes a physical storage layout of the data-intensive computer and provides access and manipulation methods for user-defined programming objects.

13. The data-intensive computer of claim 1 wherein the system management layer maintains a resource map describing storage and server resources available in the database sub-system and provides a caching system for grouping I/O requests from the application and a scheduler assigning the I/O requests to the plurality of database servers of the database sub-system.

14. The data-intensive computer of claim 1 further comprising a high-bandwidth network configured to connect the processing sub-system, the database sub-system, and the operating system sub-system.

15. The data-intensive computer of claim 14 further comprising a secondary network, separate from the high-bandwidth network, configured to provide access to the data-intensive computer by remote computers and wherein the plurality of processing servers includes at least 90 processing servers and the plurality of database servers includes at least 12 database servers.

16. The data-intensive computer of claim 1 wherein the collective database is in excess of a petabyte of storage.

17. A data-intensive computer comprising:

a processing sub-system formed by a plurality of processing servers;

a database sub-system formed by a plurality of database servers configured to form a collective database;

a unifying operating system environment extending across the processing sub-system and the database sub-system to coordinate operation of the plurality of processing servers and the plurality of database servers to act as components in a single data-intensive computer;

wherein the unifying operating system environment is further configured to implement a memory system for the single data-intensive computer to deliver pages between the processing sub-system and the database sub-system such that the plurality of database servers provide 0(log N) access to individual random pages of data stored within the collective database storing N pages of data; and

wherein the unifying operating system environment presents a system management layer, a data object layer, a database access layer, and a data transport layer to coordinate operation of the processing sub-system and the database sub-system.

18. The data-intensive computer of claim 17 wherein the unifying operating system environment is further configured to present the database sub-system to an application running in the data-intensive computer as a layer in a memory hierarchy of the data-intensive computer.

19. The data-intensive computer of claim 17 wherein the unifying operating system environment is configured to logically represent a specific data object stored in the database sub-system as a single data object when the specific data object is physically distributed across the plurality of database servers of the database sub-system.

Assignments (2)
CONFIRMATORY LICENSE Recorded Dec 14, 2017
From: JOHNS HOPKINS UNIVERSITY
To: NATIONAL SCIENCE FOUNDATION
Reel/Frame 044872/0581 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Dec 10, 2012
From: GIVELBERG, EDWARD; SZALAY, SANDOR
To: THE JOHNS HOPKINS UNIVERSITY
Reel/Frame 029434/0133 →
Continuity (2)
Provisional Application 61658189 · Jun 11, 2012
Related Publication 20140074904A1 · Mar 13, 2014