“The Deutsche Bank Technology Center is a structure for IT-support for the bank’s global business” - Alexander Khalukhin
We offer you a fresh issue of interviews with leading workshops at PG Day'17 Russia. Today we talked with Alexander Khalukhin , Oracle database developer with 10 years of experience. Alexander has been professionally engaged in development for more than 12 years. He spent the last four years of his career developing large projects for the Deutsche Bank technology center in Moscow.
At PG Day'17 Russia, Alexander will present an intensive practical course on diagnosing the performance of the Oracle Database .
Alexander told us about the internal initiatives of the technology center, the project he is working on, explained the difficulties of migrating large banking products to alternative data storage solutions and, of course, gave a brief announcement of the upcoming master class.
')
PG Day: Alexander, tell us a little about yourself. How long have you been working in the profession, what qualifications do you have?
Alexander : In development, at the age of 18, it was then that my first permanent job for a salary took place. The current specialization is now PL / SQL Oracle. I have been doing this on a permanent basis for more than ten years. Began, like many coming to Oracle in those years, with all kinds of rapid development environments Delphi and C ++ Builder. During his career, managed to work in different places. Currently I work at the Deutsche Bank Technology Center.
PG Day: Tell me more about what they do and what tasks they solve at the Deutsche Bank Technology Center?
Alexander : The Deutsche Bank Technology Center is an independent structure for IT support for the bank’s global business in all its fields. The department in which I work is currency transactions, the FX market.
As is known, Deutsche Bank has several development centers located in different parts of the world: in the United States, Romania, India, and the largest in Russia (Moscow and St. Petersburg). Technology Centers are developing new products, as well as supporting existing solutions.
PG Day: I know that various initiatives and research arise from time to time in the Technology Center. What are these initiatives? Do database specialists take part in them? If so, in what capacity?
Alexander : Indeed, from the business side there is a request for innovation. It often happens that business sees the picture in its own way, and IT specialists in their own way. Look for the point of application of forces, possible optimization should always be together. A business can suggest how to reduce costs, understand the non-optimality of business processes, and IT professionals can suggest how this is best done from a technology perspective. Therefore, initiatives are really welcome.
There are quite a number of cases in my domain among colleagues when they came to the business and said: “Let's give up this and that - we will save money on equipment maintenance and paying for licenses of purchased software. Let's replace it with this - it is being developed and alive. ” If there is some kind of non-optimality, which allows replacing several software products with one, then the point of application of forces will be applied to it so that everything will be more flexible and efficient.
I work in a database development team of one of the most high-loaded projects that considers risks in trade. We also work closely with the business. For example, a large-scale migration project for Exadata is underway. Such migration will significantly reduce costs, provide a foundation for further growth and solve many accumulated problems.
PG Day: Speaking of migrations and technologies. Can you explain what Exadata is, what goals do you follow with such a migration? What tasks are you trying to solve?
Alexander : Exadata is the name of the Oracle product, which is a proprietary piece of hardware. It is a strong server with specially sharpened storage and an Oracle cluster spinning on it. This is a clustered DBMS. What are the advantages and disadvantages of such a solution? The advantage is that everything is from a single developer, which means very simple support. In case of related problems between hardware and software, there is no need to break between three or four vendors: manufacturers of hardware, an operating system, drivers for storage or storage itself, a DBMS. All this is served from one source. In addition, these are high-tech solutions with very good performance. They are sharpened by using a high-performance machine as such a DBMS server.
A minus is expensive. Unfortunately, not all of it is available, not such a wide community. Although this is the same DBMS, the same Oracle RAC, only on a special hardware.
This will be my first experience with cluster databases, but in essence this is the same thing that we can find on other Oracle cluster solutions with the only difference that here this piece of hardware works with it a bit more optimized, it allows you to do all sorts of interesting things. . For example, filtering data during a query. If there is any predicate (the simplest case is SELECT * FROM tabname and some set of predicates), then filtering of extra records will occur at the storage level, they will not even reach the operating system.
PG Day: What is the benefit of migration? And what is the greatest test for you in this process? For someone, this is a rewrite code, for someone - data to migrate without downtime ...
Alexander : The first obvious benefit from migration is that we get space for further growth in productivity . Further, any big old project is like a ship that has been sailing the seas for many years, overgrown with shells below the waterline. So, this kind of architectural migration gives reason to cut off all excess scab, "shells", arranging an additional audit of who our customers are, with what systems we interact with. The peculiarity of the project in which I am currently working is a fairly old large system that interacts with several hundred other systems. Some of these systems are on the final steps of the life cycle, and some, on the contrary, are just beginning. This is such a rock in the raging sea, in which everyone is guided by you.
The main benefit is to put this “rock” in order. And the test, absolutely exactly you noticed, is a large number of external systems with which you need to interact and provide a "seamless" migration. Because it is clear that we do not have the moral or physical right to force all surrounding developers, support service employees to make changes for this migration in unison with us. Therefore, it is a very complex, large and coordinated synchronous process of preparing all systems or some groups of these systems and highlighting common usage patterns. In addition to several hundreds of external systems with which we interact (information system in a general sense), there are still the same number of applications that run on end-user machines. This is also quite a big "zoo". The tests listed are all ours.
PG Day: Is there a trend towards innovation in your Tech Center? Are your developers and engineers looking at other solutions (maybe cluster or NoSQL) or are you focused only on the Oracle platform?
Alexander : Of course, we evaluate different solutions. But it's good that our business understands that this kind of choice is always a double-edged sword. On the one hand, you get advantages, on the other - disadvantages. It is always a balance between maintaining running streamlined processes and the potential benefit of getting something improved at the cost of instability.
In our project, a historic decision was made in favor of Oracle , and we still have not regretted it at all. A sufficient amount of business logic is in the database, and we use, if not all, then a good 95% of all the features that Oracle offers as a commercial DBMS. At the same time, there are requests from the business for productivity and projects of a special kind where a full-fledged commercial operational DBMS is not needed. Where it is achievable and appropriate, we use NoSQL solutions. Unfortunately, I don’t have so much expertise in this area to tell you more about it.
PG Day: In your opinion, does the business of such a level as banking have the prospect of operating with the use of new alternative solutions? Or will there always be a place only for large commercial things like Oracle? I know that now there are precedents, new solutions appear. Many developers claim that they are able to compete with industrial commercial bases in terms of bandwidth, the number of transactions per second.
Alexander : As a joke, it is possible to meet a dinosaur. It’s quite possible that some open source software or non-relational DBMS can force out classic commercial monsters like Oracle, SQL Server, Teradata in banking. This will happen as soon as they become quite stable both in work and in a supported API, they will acquire a good layer of development and diagnostic tools, will form a sufficient community and technical support with a reasonable SLA.
As I understand from my experience, the banks have a sufficient amount of legacy software to ensure their basic livelihoods. The transition from it to something more modern, with better performance and functionality is complicated not so much technologically as organizationally. It would seem that nothing costs us to migrate the data, to write a new software for interacting with this data. But de facto we will face the fact that we will have to retrain users, technical support service, we will need to take into account the immense number of systems that interact with you. Technically, everything is feasible, but in practice it is work that is outside the developer’s domain.
Some new projects are already using NoSQL solutions as an alternative to standard relational DBMS.
PG Day: Continuing to talk about your expertise in Oracle as a DBMS: what features does Oracle have that differentiate it from other data storage solutions?
Alexander : Top features that I admire at Oracle are his excellent built-in diagnostic tools , this is the amount of information that he gives you about what is happening in it. Due to the fact that quite a lot of time has to be dealt with not only developing, but also supporting existing solutions, debugging performance, optimization, I don’t know what I would do in the absence of these tools. So, for example, as far as I know, in PostgreSQL this is a rather acute problem, the infrastructure of diagnostic tools is not well developed there.
In Oracle, this way: if there is a task, it means that there is a standard tool for it. If there is no standard instrument, you can make a composition of several other instruments in order to achieve the same result. I am not so much a fan of new functionality, which he is also famous for, but rather of how well he is instrumented.
PG Day: What do you think can be improved in Oracle?
Alexander : Here I will not be original - the quality of the support itself . Since I work in a project that uses a large number of different technologies, I had the opportunity to compare the support of many developers. I am still very impressed with the support provided by DELL. This is the company that bought QUEST, which it had previously made known to many Toad, SQL Navigator, Shareplex (a product for replicating data from the times “prior to purchasing GoldenGate Oracle”). Among all the development companies that I managed to encounter, DELL, or at least that part of the team that supports Quest products, is the standard for IT support. These are very responsive people.
They can call at any time of the day or night, and they will find an engineer who will solve the problem with you before the victorious. Unfortunately, I do not see such zeal, such a good quality of support services provided by Oracle. It plunges me into some disheartening, because I know that the bank is paying quite a lot of money for licenses. This is not our unique problem, not only our “headache”. I think that many "Oraklists" who have ever had to interact with Oracle-support have the same complaints. If you succeed in reaching a second or third level engineer who will make a start from typical questions and try to offer something, then good support begins with this. But, unfortunately, in order to achieve this, an incredible amount of diagnostics and other things will “shake off” you, which will have nothing to do with the problem you are describing.
PG Day: Tell us a little about the master class that you will hold at the conference in the summer on PG Day. It is dedicated to diagnosing the performance of Oracle Database. What questions do you plan to consider, what were you guided by when you made the plan for the master class?
Alexander : The master class will be devoted to the issues of diagnostics, first of all, diagnostics of performance problems . Among the whole range of problems that developers and support staff have to face, they are the least obvious. With functional problems it is clearer: there is a bug, found a problem in the code and fixed it. When a problem flashes or it happens in the past, we may not have enough diagnostics for this, and we were not sufficiently prepared.
At the meeting, I will talk about everything that is a common problem in other DBMSs and what I admire about Oracle. I will build my story on the basis of personal experience and my own approaches in diagnostics.
I will try to present everything in the form of a lively interactive, which will not only lift the veil into the world of Oracle's diagnostic capabilities, but also provide quite concrete practical experience in using these diagnostic tools to overcome problems. First of all, I’ll touch on what consumes the most time for a developer and a supporter. I'll tell you how to diagnose problems in Oracle and deal with them . What patterns and approaches in diagnostics can be used and transposed from Oracle to PostgreSQL, ideologically closest to Oracle in terms of the DBMS architecture.
PG Day: Thank you, Alexander!
At PG Day'17 Russia, Alexander will present an intensive practical course on diagnosing the performance of the Oracle Database .
Alexander told us about the internal initiatives of the technology center, the project he is working on, explained the difficulties of migrating large banking products to alternative data storage solutions and, of course, gave a brief announcement of the upcoming master class.
')
PG Day: Alexander, tell us a little about yourself. How long have you been working in the profession, what qualifications do you have?Alexander : In development, at the age of 18, it was then that my first permanent job for a salary took place. The current specialization is now PL / SQL Oracle. I have been doing this on a permanent basis for more than ten years. Began, like many coming to Oracle in those years, with all kinds of rapid development environments Delphi and C ++ Builder. During his career, managed to work in different places. Currently I work at the Deutsche Bank Technology Center.
PG Day: Tell me more about what they do and what tasks they solve at the Deutsche Bank Technology Center?
Alexander : The Deutsche Bank Technology Center is an independent structure for IT support for the bank’s global business in all its fields. The department in which I work is currency transactions, the FX market.
As is known, Deutsche Bank has several development centers located in different parts of the world: in the United States, Romania, India, and the largest in Russia (Moscow and St. Petersburg). Technology Centers are developing new products, as well as supporting existing solutions.
PG Day: I know that various initiatives and research arise from time to time in the Technology Center. What are these initiatives? Do database specialists take part in them? If so, in what capacity?
Alexander : Indeed, from the business side there is a request for innovation. It often happens that business sees the picture in its own way, and IT specialists in their own way. Look for the point of application of forces, possible optimization should always be together. A business can suggest how to reduce costs, understand the non-optimality of business processes, and IT professionals can suggest how this is best done from a technology perspective. Therefore, initiatives are really welcome.
There are quite a number of cases in my domain among colleagues when they came to the business and said: “Let's give up this and that - we will save money on equipment maintenance and paying for licenses of purchased software. Let's replace it with this - it is being developed and alive. ” If there is some kind of non-optimality, which allows replacing several software products with one, then the point of application of forces will be applied to it so that everything will be more flexible and efficient.
I work in a database development team of one of the most high-loaded projects that considers risks in trade. We also work closely with the business. For example, a large-scale migration project for Exadata is underway. Such migration will significantly reduce costs, provide a foundation for further growth and solve many accumulated problems.
PG Day: Speaking of migrations and technologies. Can you explain what Exadata is, what goals do you follow with such a migration? What tasks are you trying to solve?
Alexander : Exadata is the name of the Oracle product, which is a proprietary piece of hardware. It is a strong server with specially sharpened storage and an Oracle cluster spinning on it. This is a clustered DBMS. What are the advantages and disadvantages of such a solution? The advantage is that everything is from a single developer, which means very simple support. In case of related problems between hardware and software, there is no need to break between three or four vendors: manufacturers of hardware, an operating system, drivers for storage or storage itself, a DBMS. All this is served from one source. In addition, these are high-tech solutions with very good performance. They are sharpened by using a high-performance machine as such a DBMS server.
A minus is expensive. Unfortunately, not all of it is available, not such a wide community. Although this is the same DBMS, the same Oracle RAC, only on a special hardware.
This will be my first experience with cluster databases, but in essence this is the same thing that we can find on other Oracle cluster solutions with the only difference that here this piece of hardware works with it a bit more optimized, it allows you to do all sorts of interesting things. . For example, filtering data during a query. If there is any predicate (the simplest case is SELECT * FROM tabname and some set of predicates), then filtering of extra records will occur at the storage level, they will not even reach the operating system.
PG Day: What is the benefit of migration? And what is the greatest test for you in this process? For someone, this is a rewrite code, for someone - data to migrate without downtime ...
Alexander : The first obvious benefit from migration is that we get space for further growth in productivity . Further, any big old project is like a ship that has been sailing the seas for many years, overgrown with shells below the waterline. So, this kind of architectural migration gives reason to cut off all excess scab, "shells", arranging an additional audit of who our customers are, with what systems we interact with. The peculiarity of the project in which I am currently working is a fairly old large system that interacts with several hundred other systems. Some of these systems are on the final steps of the life cycle, and some, on the contrary, are just beginning. This is such a rock in the raging sea, in which everyone is guided by you.
The main benefit is to put this “rock” in order. And the test, absolutely exactly you noticed, is a large number of external systems with which you need to interact and provide a "seamless" migration. Because it is clear that we do not have the moral or physical right to force all surrounding developers, support service employees to make changes for this migration in unison with us. Therefore, it is a very complex, large and coordinated synchronous process of preparing all systems or some groups of these systems and highlighting common usage patterns. In addition to several hundreds of external systems with which we interact (information system in a general sense), there are still the same number of applications that run on end-user machines. This is also quite a big "zoo". The tests listed are all ours.
PG Day: Is there a trend towards innovation in your Tech Center? Are your developers and engineers looking at other solutions (maybe cluster or NoSQL) or are you focused only on the Oracle platform?
Alexander : Of course, we evaluate different solutions. But it's good that our business understands that this kind of choice is always a double-edged sword. On the one hand, you get advantages, on the other - disadvantages. It is always a balance between maintaining running streamlined processes and the potential benefit of getting something improved at the cost of instability.
In our project, a historic decision was made in favor of Oracle , and we still have not regretted it at all. A sufficient amount of business logic is in the database, and we use, if not all, then a good 95% of all the features that Oracle offers as a commercial DBMS. At the same time, there are requests from the business for productivity and projects of a special kind where a full-fledged commercial operational DBMS is not needed. Where it is achievable and appropriate, we use NoSQL solutions. Unfortunately, I don’t have so much expertise in this area to tell you more about it.
PG Day: In your opinion, does the business of such a level as banking have the prospect of operating with the use of new alternative solutions? Or will there always be a place only for large commercial things like Oracle? I know that now there are precedents, new solutions appear. Many developers claim that they are able to compete with industrial commercial bases in terms of bandwidth, the number of transactions per second.
Alexander : As a joke, it is possible to meet a dinosaur. It’s quite possible that some open source software or non-relational DBMS can force out classic commercial monsters like Oracle, SQL Server, Teradata in banking. This will happen as soon as they become quite stable both in work and in a supported API, they will acquire a good layer of development and diagnostic tools, will form a sufficient community and technical support with a reasonable SLA.
As I understand from my experience, the banks have a sufficient amount of legacy software to ensure their basic livelihoods. The transition from it to something more modern, with better performance and functionality is complicated not so much technologically as organizationally. It would seem that nothing costs us to migrate the data, to write a new software for interacting with this data. But de facto we will face the fact that we will have to retrain users, technical support service, we will need to take into account the immense number of systems that interact with you. Technically, everything is feasible, but in practice it is work that is outside the developer’s domain.
Some new projects are already using NoSQL solutions as an alternative to standard relational DBMS.
PG Day: Continuing to talk about your expertise in Oracle as a DBMS: what features does Oracle have that differentiate it from other data storage solutions?
Alexander : Top features that I admire at Oracle are his excellent built-in diagnostic tools , this is the amount of information that he gives you about what is happening in it. Due to the fact that quite a lot of time has to be dealt with not only developing, but also supporting existing solutions, debugging performance, optimization, I don’t know what I would do in the absence of these tools. So, for example, as far as I know, in PostgreSQL this is a rather acute problem, the infrastructure of diagnostic tools is not well developed there.
In Oracle, this way: if there is a task, it means that there is a standard tool for it. If there is no standard instrument, you can make a composition of several other instruments in order to achieve the same result. I am not so much a fan of new functionality, which he is also famous for, but rather of how well he is instrumented.
PG Day: What do you think can be improved in Oracle?
Alexander : Here I will not be original - the quality of the support itself . Since I work in a project that uses a large number of different technologies, I had the opportunity to compare the support of many developers. I am still very impressed with the support provided by DELL. This is the company that bought QUEST, which it had previously made known to many Toad, SQL Navigator, Shareplex (a product for replicating data from the times “prior to purchasing GoldenGate Oracle”). Among all the development companies that I managed to encounter, DELL, or at least that part of the team that supports Quest products, is the standard for IT support. These are very responsive people.
They can call at any time of the day or night, and they will find an engineer who will solve the problem with you before the victorious. Unfortunately, I do not see such zeal, such a good quality of support services provided by Oracle. It plunges me into some disheartening, because I know that the bank is paying quite a lot of money for licenses. This is not our unique problem, not only our “headache”. I think that many "Oraklists" who have ever had to interact with Oracle-support have the same complaints. If you succeed in reaching a second or third level engineer who will make a start from typical questions and try to offer something, then good support begins with this. But, unfortunately, in order to achieve this, an incredible amount of diagnostics and other things will “shake off” you, which will have nothing to do with the problem you are describing.
PG Day: Tell us a little about the master class that you will hold at the conference in the summer on PG Day. It is dedicated to diagnosing the performance of Oracle Database. What questions do you plan to consider, what were you guided by when you made the plan for the master class?
Alexander : The master class will be devoted to the issues of diagnostics, first of all, diagnostics of performance problems . Among the whole range of problems that developers and support staff have to face, they are the least obvious. With functional problems it is clearer: there is a bug, found a problem in the code and fixed it. When a problem flashes or it happens in the past, we may not have enough diagnostics for this, and we were not sufficiently prepared.
At the meeting, I will talk about everything that is a common problem in other DBMSs and what I admire about Oracle. I will build my story on the basis of personal experience and my own approaches in diagnostics.
I will try to present everything in the form of a lively interactive, which will not only lift the veil into the world of Oracle's diagnostic capabilities, but also provide quite concrete practical experience in using these diagnostic tools to overcome problems. First of all, I’ll touch on what consumes the most time for a developer and a supporter. I'll tell you how to diagnose problems in Oracle and deal with them . What patterns and approaches in diagnostics can be used and transposed from Oracle to PostgreSQL, ideologically closest to Oracle in terms of the DBMS architecture.
PG Day: Thank you, Alexander!
Source: https://habr.com/ru/post/331150/
All Articles