DE10046614C1 - Electronic engineering application execution method for computing centre has user coupled to portal server of computing centre - Google Patents

Abstract

The electronic engineering application execution method has a user (A) coupled to a portal server (PS) of the computing centre, e.g. an internet computing centre (ICC1) via an encoded coupling (VV), with the data for the performed calculation supplied to the user layer (AL) of the portal server upon completion of the calculation, with indication of the latter for the user.

Description

The invention relates to a method for executing electronic Engineering applications in a computing center with a portal server via the multiple users can submit computing jobs to the Computing Center with one Database server on which the data of the user and his orders are stored and with a compute server on which the orders are calculated.

In the simulation and calculation areas of the R&D departments of the vehicle, Mechanical, aerospace, telecommunications and semiconductor industries Large computers are used to measure the physical and logical properties of end products (e.g. cars, semiconductors, airplanes) or their components simulate. A simulation is used to predict the behavior of the products, for example their lifespan, stability, security and their consumption. On Calculation engineer uses either program packages available on the market or a self-developed simulation software. This is done by the company Provided computers installed and licensed. The computer itself is in one Company network installed. The company is for the selection (Benchmarking), procurement (e.g. leasing or purchase), installation, provision of usually air-conditioned rooms and operation, for example (by a Service provider or on your own responsibility). The calculation engineers and researchers access these computing resources through simple Command line interfaces or with the help of graphical user interfaces. A however, such an approach has some disadvantages for the calculation engineer or the company with itself. Procuring a calculator is an investment It only pays off if the computer has a good to very good lifetime is busy. After the computer speed doubled approximately every 18 months, however, a two year old calculator is no longer acceptable for many calculations, cannot, however, be easily replaced, for example due to depreciation become. In addition, buying the resources as a calculator means that there is no flexibility with regard to a possibly higher required computing power. The Company is responsible for the service itself or awards it to someone external operator. By separating procurement and operation Friction losses. In a service level agreement (SLA) only technical  Metrics are applied, however, is not applied to the company's business processes received. Due to increasing globalization and worldwide development in Businesses need resources 24 hours a day, regardless of location be usable. However, this requires a high level of security technology. There even within a company the command line based access or the Access with simple graphic interfaces is not very standardized expensive maintenance, high expenses for the training of new employees and a Depends on the operating system.

DE 196 37 219 A1 describes an apparatus and a method for detecting Processing and evaluation of medical measurement data with a via a dial-up network a remote computer connected local computer known. The device includes at least one examination terminal that connects to a suitable interface the local computer is connected, the control of at least one Examination terminals and the implementation and recording of the measured values the remote computer is done. In the process, the remote computer is activated by the Local computer selected to establish a connection, the remote computer is switched on a corresponding examination terminal and the control function of the Examination terminals are taken over. Furthermore, a measurement medical data taking into account predetermined measurement parameters as well as the measurement data evaluated in the remote computer and the evaluation results in the Remote computers provided. However, it cannot be guaranteed that the highly sensitive medical data of patients are protected to the extent that only the the doctor treating them can read them.

In US 5,754,830 a computer network environment is described in which one Connection between a client system and one for checking Authorizations responsible central computer via a server. The The central computer has a TCP / IP connection. The central computer supports a terminal Session when accessing this. A client connection communicates with the Central computer via a permanent TCP / IP socket connection. In the client system an application process is executed under a web browser. The application Process communicates with the client connection through another permanent TCP / IP Socket connection and provides a terminal session to a user of the client Systems available. The terminal session is through a constant connection supports bidirectional communication in real time with the central computer  allows. Here too, no transmission of the data is guaranteed, with which this only that user can recognize for whom they are intended.

US 5,893,905 describes a system and a method for monitoring the quality a selected data processing operation. This happens because the current quality with the Service Level Agreement (SLA) of everyone monitored Order is compared and deviations are identified as well as effects be analyzed for other jobs in a job stream. If a delay is one Jobs is detected, this is communicated to the user and the critical path of the SLA identified. The impact on the SLA of dependent jobs is determined. The report to the user is done automatically when there is a risk that the SLA will not can be achieved. Deviations are also displayed, for example for whom the job runs faster or slower than normal. This also results here Problem that no secure transmission of sensitive data is only for one certain users should be accessible is possible.

EP 0 268 141 A2 describes a terminal that is physically and logically secured processor. An encrypted decryption key is transferred from an authorization diskette to the processor. The terminal then writes an encrypted file on the authorization diskette to itself Identify as the correct terminal and get authorization. The disk is then returned to the computer center to check if the correct one Processor has received authorization. The decryption key is used for the Use of the central computer is required. After the user of the terminal his Identification passed on to the central computer, the whole is subsequent Communication between the terminal and the central computer is encrypted and can only be done by the decryption code that is in the co-processor is decrypted. The decryption code can be dynamic for more Transfers are changed. A transfer of the decryption key to the The co-processor can also be used without an authorization diskette. To do this, a special program integrated in the co-processor, which can be used to to determine with certainty whether the correct terminal is authorized. This is a very complicated procedure for the safe and secret transmission of sensitive data to ensure.

The object of the present invention is therefore to access data and Ensure applications with a high level of security, both based on Measures are taken on the user side as well as on the server side to achieve a To ensure authenticity.

This object is achieved by a method with the features of claim 1. Advantageous developments of the invention are the subject of the dependent claims.

In the method according to the invention, engineers and researchers use a portal Possibility to access engineering applications. On their To do this, they receive an applet or another access system on the desktop. The portal is a Part of the customer's process chain: He puts data into the portal, specifies it Parameters and boundary conditions and determines the time of the calculation. The Portal bundles various functions and provides the user with the actual one Simulation offers a wide range of analysis options: In addition to the Job status will also visualize the partial results during and after Simulation end enabled. For example, the course of a crash simulation be observed during the calculation. The portal standardizes access to the Resources and provides the user with an overview of costs, one Electricity meter comparable, available. This is customer-specific accounting an integral part of the portal. The data is sent to one via the portal Portal server passed in a computing center. Having such a portal can use multiple users who are not necessarily from one company According to the invention, it is provided that for reasons of security and secrecy all data belonging to an order only to the Users who placed the order are accessible. This data will encrypted. It is also envisaged that after the end of the  Computing job this is displayed. According to the invention, after the Calculating an order the data of this computing order on the application layer given the portal server. It is through the aforementioned method according to the invention therefore only possible for the user of a computing job, information and data to get over his arithmetic order. Authenticity is thus guaranteed.

An advantageous development of the invention provides that the computing Center is an Internet Computing Center and the encrypted connection and the indication that the computing job has ended is made via the Internet. Thereby there is the possibility for the user to access the job data both via the normal Desktop in the company as well as via a mobile client, for example one Laptop or a handheld, to be able to access via the Internet.

It is also advantageous if the results are automatically, especially via the Internet. With this, the user immediately gets that The result of his computing job without him constantly asking the computing center whether the computing job has already been completed.

It is also advantageous if the portal server contains the user data of each user compares with a user database and checks whether the user has a credit rating for has the registered process, the billing based on the data done by the database server. On the one hand, this serves to increase security, since the Authenticity of the user can be checked and at the same time to ensure that the The operator of the Computing Center ensures that the user orders it given arithmetic order is also paid.

It is also advantageous if the Computing Center with other Computing Centers is connected to which it can redirect a user's computing task and from which it can receive computing jobs, depending on the workload of the individual computing Center. This makes it possible for the Computing Center to use its own workload optimally achieved, whereby it compensates for peaks in the own load of its CPUs can and an overbooking is avoided, which leads to a waiting time for one Billing order would lead. This also makes it possible for the capacity of the Computing center does not have to be designed so high that a peak during peak times immediate completion of the computing orders is possible. There will be enormous costs in Computing center saved.  

It is also advantageous if the user spends computing time in the computing center can reserve and a current computing job in the Computing Center if there is sufficient computing time available less the reserved computing time stands and the computing job is otherwise forwarded to another computing center that currently has enough computing time, with the coordination of one Decision service is made, which is for the allocation and sending relevant data of all connected computing centers. This makes it possible that computing time booked in advance by a user, for example up to can be booked three months in advance (the so-called Long Term Scheduling (LTS)), can be taken into account. This will increase the predictability of Computing center reached, which enables more efficient resource utilization, which ultimately has an impact on the price. After the Computing Center over Having the data of the other computing center is an optimal utilization of the entire network of all connected computing centers possible.

Furthermore, it is advantageous if when evaluating whether and to what computing Center will pass an order, the availability of CPU time in its own Computing Center, the cost of transferring data to the particular computing Center, the computing costs there, the utilization of the respective computing center as well the job values for competing orders are taken into account. This will make one Cost optimization achieved because it includes all the costs for the user, is possible, the most convenient way for all possible ways to choose. By specifying job values, an optimal Control of the distribution of computing jobs to all computing centers achieved become.

In the following, further advantageous refinements of the invention are shown in the following figures described. Here show:

Fig. 1 is a block diagram showing a flow of a method according to the invention,

Fig. 2 is a schematic view of an LTS and

Fig. 3 is a flow diagram illustrating the verification of the creditworthiness and a scheduling.

The method shown in FIG. 1 is principally characterized by two steps. First, the workflow from a user A into a global system GS from several Internet Computing Centers (ICC), one component being a Compute Center CC in an intranet, and second, the workflow within a single ICC. A calculation order is placed anywhere in the system, the individual parts of which are connected to one another via the Internet. The sequence of the method described below then results there.

In a first step 1, user A selects the URL of a portal server PS on the Internet. The connection established is encrypted using a server-side certificate, preferably an X.509 certificate.

After the encrypted connection W has been established, the portal server PS sends the applicant A a registration form in a second step 2 .

In a third step 3 , user A registers with the portal server PS with a user ID and a password.

In a fourth step 4, the portal server PS checks the user ID and the password against a user database BD. Further details of the check are described with reference to FIG. 3.

Depending on his rights, in a fifth step S the user A now actual application frontend made available in the form of an HTML page or the request is rejected if user A has no authorization.

If user A has authorization, he formulates his request in a sixth step 6 and sends it over the Internet to portal server PS.

In a seventh step 7, the portal server PS interprets the request and forwards it to a decision service DS, which searches for suitable resources for the required computing job in accordance with the specification of user A, compiles and sends the job for it. More on this will be explained with reference to FIG. 3. The decision service DS receives the data for the decision 7 a from a database server DBS1 of a first Internet computing center ICC1. This first database server DBS1 is connected via a connection to further database servers DBS2, DBS3 of several Internet computing centers ICC2, ICC3. This connection is used to exchange resource information 7 b via a database protocol. The corresponding input data is also transmitted beforehand using a secure protocol. The two Internet Computing Centers shown are in no way restrictive; it is equally possible that three, four or more Internet Computing Centers are connected.

In an eighth step 8 , the information about the submitting job, for example ID, user name, company, department, billing account, limit, priority, job status, is stored in the first database server DBS1 for later reuse for accounting / billing (pay-per-use). According to the invention, the job status can be called up by user A at any time, which also includes billing information, but it is not possible for user A to be able to access data from another job.

In a ninth step 9 , the decision service DS sends the calculation job to a local first job switch JS1.

In a tenth step 10 , the first job switch JS1 advantageously receives information about whether the calculation job is to be carried out and either forwards it directly to a first local compute server CS1 via a connection 10 a or it commissions a job switch of a non-local job switch JS2, JS3 to forward the job to a compute server CS2, CS3 of the respective Internet computing center ICC2, ICC3. This is done via further connections 10 b, 10 c.

In an eleventh step 11 , the results of the calculation are advantageously transmitted to the portal server PS via connections 11 a, 11 b, 11 c. There you go to the application layer AL.

In a twelfth step 12 the portal server PS is informed after the end of the computing job, which then updates the information in the first database server DBS1.

According to the invention, in a thirteenth step 13, user A is informed of the termination of his computing job. This is advantageously done via the Internet by email. At the same time, the result A can be transmitted to user A at his request. After this is done according to the invention via the encrypted connection VV, a high security standard is guaranteed. Alternatively, the user can selectively retrieve the result data through the portal.

A scheduling is shown schematically in FIG . For optimal utilization of computer resources, it is advantageous for a provider if he knows in advance when and how much CPU time must be made available. He can then plan and inquire at another Internet computing center ICC2, ICC3 whether they can provide CPU time at peak times when he is overbooked 14 . A long-term scheduling (LTS) enables a customer to make a reservation request for a long and medium period of time. For example, he can reserve 50 CPU hours of a certain type with defined software and memory requirements for the next 3 months. He can also determine the priority distribution, for example 50% of all jobs have the highest priority, 50% only normal priority. The system is also so flexible that it can reserve more or less CPU time at certain times. An LTS reservation has tangible economic effects for customers, and the price is lower due to the optimal use of computer resources. In the example shown, a first customer K1 and a second customer K2 have made LTS reservations for the next few months, with only a maximum of CPU hours being available. The first customer K1 made LTS reservations throughout the entire period. In contrast, the second customer K2 only reserved certain time slots. Other time windows 15 a, 15 b, 15 c, 15 d are still free. During the period T between the two free time windows 15 b, 15 c at the second customer K2, however, an overbooking 14 is given. An incoming for this period T arithmetic order must be given either in a holding pattern and can then during the next available time 15 c of the second customer K2 are inserted, the calculation order is not necessarily the second customer K2 are possible needs, but also from any other customer is. Or the computing order of the overbooking 14 is transferred to another Internet computing center ICC2, ICC3, which still has computer resources available at the time in question. Further details are described with reference to FIG. 3.

FIG. 3 shows the process sequence when checking the creditworthiness of a user A and the distribution of the computing orders to different Internet computing centers. After transmission 16 of the job data, the access permission of user A is checked. When checking 17 , the user ID and the password are compared with the user database BD (see fourth step 4 in FIG. 1). If it turns out that user A has no access authorization, the job is ended. If the authorization has been successfully completed, a check 18 of the account balance of user A is carried out. If the account is empty, an additional check is made to determine whether an overdraft is possible. If this check 19 shows that the account is empty, the job is canceled. Otherwise, the determination 20 of a job value is made in the same way as after a positive account check. This is done by analyzing the job parameters entered. The job value is determined from the economic benefits, such as the value of the CPU hours, licenses and priorities. This value is crucial for the Decision Service DS.

The job profile determined is used to check the extent to which the request can be met using the locally available resources of the Compute Server CS1. The availability check 21 shows whether local scheduling (job submission) can take place. This is known as Short Term Scheduling (STS). In addition to the static resource requests, such as hardware architecture and software type, as well as software versions, the STS also uses dynamic variables, such as CPU usage, memory usage and network performance, for distribution. A connection 22 to the database server DBS1 also uses the medium and long-term reservations 23 by means of LTS. SVA information 24 , account information 25 and ICC resources 26 also flow into the control process via the database server DBS1. The LTS provides information about the utilization of the respective local ICC resources within the next 3 months. These must be taken into account at the STS.

If the resources are available and a request via LTS is not high Loads can be expected within the expected CPU time, the Calculation job started on the selected resources. In case it is competing Jobs, a decision as to which job is started first is made using the Evaluation of the job value taken. If the local Internet computing center ICC1 If there are enough resources available, the job data is transferred to it.  

The job is then executed 27 . A transfer 28 of the results obtained to the application layer AL of the portal server PS follows and the job 29 has been ended.

If the resources of the local Internet computing center ICC1 are not sufficient, then the job was forwarded to a remote Internet Computing Center ICC2. There will be then the steps just mentioned for the local ICC are processed. After a transfer the job data to the remote Internet computing center ICC2 will be used there executed and the result is transferred back to the Portal Server PS. The job is with it completed.

The job value is to be rated higher than the priority requested by the user. at the same job value decides the priority. If the resources are available the LTS database, however, suggests that resource utilization is expected, the job is only started if the job value is higher than that of the expected jobs. If not, the job has to be put on hold. Every job in STS the LTS service must agree in advance.

A portal system is advantageously constructed from individual components, for their Realization following technologies are favored: The user front end is with HTML and JavaScript. With complex graphical interfaces, Java applets are used used. An Apache SSL with a servlet engine, for example, is used as the portal server the Jserv or tomcat. X.509 certificates are created with openssl. LSF is used as job management and workload management system. JDBC is used as the interface to the databases. As databases first used MySQL and / or Informix. The programming language is Java. As Communication middleware between the applications and LSF becomes remote Method Invocation (RMI) used.

Using a commercial application server results in the above the following differences: It becomes a proprietary portal server with its own Servlet engine and own SSL implementation used. For communication RMI or EJB used. Instead of the databases mentioned above you can also supplied databases can be used.  

The computing power on the Internet using the portal is advantageously the Offered to customers based on SVAs. These include, in addition to those offered Services also the description of the customer process and the most accurate Description of the influences on this process by using the portal in Form of a metric. In traditional metrics there is a correlation between success of the business process and service level agreements are difficult or impossible to identify. Therefore, it makes sense to use metrics between the business process-oriented and the distinguish technical metrics and summarize them as performance metrics describe.

The business process-oriented measures include: deviation of the target / actual Project duration, speed of resource deployment, and speed at Provision of new software and licensing.

Technical metrics include: availability of resources, for example Software, hardware, networks, email and turn-around times of batch processes.

The services of the provider are usually based on the measured values monthly or recorded quarterly. To get a proactive approach, are however, these periods are too long. The critical success factors must be weekly or measured daily so that timely measures can be initiated. By This ensures constant access to the data by the user. Are the Measured variables can be recorded automatically, so software tools should be used for permanent monitoring can be used, which allow these tasks to be carried out without major personnel expenditure cover.  

LIST OF REFERENCE NUMBERS

1-13

Step carried out

7

a decision

7

b exchange

10

a connection

10

b connection

10

c connection

11

a connection

11

b connection

11

c connection

14

overbooking

15

a Free time window

15

b Free time window

15

c Free time window

15

d Free time window

16

transmission

17

Verification

18

Verification

19

verification

20

Determination of the job value

21

Check whether local resources are available

22

Connection to the database server

23

Reservations

24

SVA Information

25

Account Information

26

ICC resources

27

execution

28

transmission

29

termination
A user
AL application layer
BD1 user database

1

BD2 user database

2

BD3 user database

3

CC Compute Center
CS1 Compute server

1

CS2 Compute server

2

CS3 Compute server

3

DS Decision Service
GS global system
ICC1 Internet Computing Center

1

ICC2 Internet Computing Center

2

ICC3 Internet Computing Center

3

JS1 job switch

1

JS2 job switch

2

JS3 job switch

3

PS Portal Server
T duration
VV encrypted connection

Claims (7)

1. Method for executing electronic engineering applications in a computing center (CC) with a portal server (PS) via which several users (A) can issue computing jobs to the computing center (CC) with a database server (DBS1) on which the data of the user (A) and his orders are stored and with a compute server (CS1) on which the orders are calculated, characterized in that the users (A) each with the portal server (PS) via an encrypted connection (VV) are in contact, the data of the computing job are given to the application layer (AL) of the portal server (PS) after completion of the calculation and all data on all jobs of a user (A) can only be viewed by the user and the user ( A) the end of his computing job is displayed. 2. The method according to the preceding claim, characterized in that the Computing Center (CC) is an Internet Computing Center (ICC1) acts as well as the encrypted connection (VV) and the display, that the computing job has ended is done over the Internet. 3. The method according to any one of the preceding claims, characterized in that the results to the user (A) automatically, especially via the Internet, to be communicated. 4. The method according to any one of the preceding claims, characterized in that the portal server (PS) contains the user data of each user (A) User database (BD) compares and checks whether the user (A) has one Has creditworthiness for the registered process, whereby the Invoicing based on data from the database server (DBS1) he follows.   5. The method according to any one of the preceding claims, characterized in that the Computing Center (CC) with other Computing Centers (ICC2, ICC3) is connected to which it can redirect a computing request from a user (A) can and from which it can receive computing orders, depending on the utilization of the individual computing center. 6. The method according to any one of the preceding claims, characterized in that the user (A) reserve computing time in the Computing Center (CC) can leave and a current computing job in the Computing Center (CC) is done if enough less the reserved computing time Computing time is available, and the computing job otherwise at one other Computing Center (ICC2, ICC3) is currently being forwarded has enough computing time, the coordination of a decision Service (DS) is made, which is responsible for the occupancy and the Send relevant data from all connected computing centers (ICC1, ICC2, ICC3) knows. 7. The method according to the preceding claim, characterized in that in assessing whether and to which computing center (ICC1, ICC2, ICC3) an order will pass on the availability of CPU time in the own computing center (ICC1), the cost of data transmission the respective computing center (ICC2, ICC3), the computing costs there, the Utilization of the respective Computing Center (ICC1, ICC2 ICC3) and the Job value for competing orders are taken into account.