Google vice President leaked mystery: Google search Technology Secrets

Google vice President leaked mystery: Google search technology uncovered--

Google owes its success in part to the company's unique computing mechanism, which allows its search engines to provide answers to the search requests of millions of web users in a very short time.
In Wednesday, Google Vice President Urs Hoelzle a speech at the Eclipse meeting to give attendees an opportunity to understand how Google's search technology is being produced and how the technology is working now.
To invent Google's technology, Hoelzle told attendees, developers must abandon the mindset of the technology used by large databases in the past. Since the content of all search requests is unpredictable in one day, it is a real challenge to have 10 billion or so pages of data on hand.
Hoelzle produced a series of photos of the early Google Hardware data center. The photos taken in 1997 were two worn-out desktop computers. 1999, is a number of Intel servers, behind the machine exposed a pile of messy data lines. In the 2000 data Center, 1000 neatly arranged dual-processor servers were placed.
"The underlying hardware is very inexpensive, but it has done a lot of work," Hoelzle said. "At the same time, using so many servers, reliability is another focus of Google's concerns." "Google operates with an automatic control mechanism," Hoelzle said. Otherwise, engineers need to be on the down and out of the server. "
In order to resist the impact of force majeure, Google built the Google file system, which is closely integrated with Google's search operation system, and has a high capacity to withstand server failures.
All of Google's operations are based on a series of large-capacity files that are split into 64M-sized packets and distributed across multiple packet servers. Information such as the description of the file, the number of packets, and the location of the packet are stored in the central server. Each 64M packet is backed up on two other servers, and three copies of the path are stored on the central server.
Because all of the files are stored on the Red Hat Linux server, Google guarantees the reliability of the service at a very low cost. The central server periodically sends a pulse signal to the packet server to determine whether the packet server is functioning properly. If no response signal is received, or if the data in the answer signal indicates that a packet server is compromised, the central server takes out the packet to repair the compromised server from the other packet servers. This work is usually done in a minute.
Hoelzle points out that only a copy of the three servers would be compromised, which would have an impact on Google's services. At this point, it takes a long time to collect data from the Internet to reconstruct the damaged files.