Books and articles about SQL Rambler's Top100 : 24 April 2014 20:35:43
Skip Navigation Links  


Print  Print version

Microsoft's employee about SQL-EX.RU

Alexey Shulenin,
subject matter expert (SQL Server & BI),
evangelist, DPE, Microsoft

Being an open educational research source, the site SQL Exercises ( is quite a valuable educational and methodical project, enabling within limited period of time to develop the skills of SQL use up to a master-class level. Notwithstanding the fact that the number of distance learning systems is exponentially growing during the recent period of time, I would characterize this project not only as innovative by nature, but also as unique due to its following characteristics.

1. The chosen field of activity.

1.1. Software development embraces numerous fields beginning with design of user interface and up to system programming. Nowadays the majority of programmers in the world are applied specialists, focused on the aspect of businesses logics. With few exceptions, there is no such thing as pure applied specialist. For example, having no idea, how transactions operate on the level of DBMS, it would be rather problematic, I suppose, to operate them on the level EnterpriseServices / +. How can one expect to model the scalable application, not knowing the principles of blocking and isolation of transactions? How can one submit the necessary database data on the form or to determine the component of business logics, having no knowledge of SQL? Thus, each developer of business logics is bound to be to a certain extent also a database developer, and thus the content of site will be useful for numerous developers.

1.2. I have come across research of Gartner, in which the technological priorities of development of IT were determined, the results being based on the inquiry of 1400 managers of information departments all over the world: the key question was what were the major fields your company is going to invest in the nearest future and what will be the proportional share of such investment ( As it is clear from the results of the report, the mysterious BI abbreviation is number one in the rating. This term was introduced in 1958 (, and initially its meaning was similar to that currently associated with DSS notion (decision support systems). Later on it acquired a more general meaning. Currently BI (Business Intelligence) signifies a set of technology tools focused on querying, reporting and multi-dimensional analysis on top of a data warehouse. Data warehouses are traditionally supplied with object-relational) database servers, and thereby SQL is used as means of communication with such servers. Teaching SQL, site ultimately helps people to acquire profession that according to the forecasts is expected to be in high demand.

1.3. CIO-priorities are reinforced by the development of the market of BI-applications on the whole (, the amount of growth in this sphere amounted to 11.5% in 2006 and to 12.5% in 2007. It is expected that this rate will steadily increase up to 2011, when the total market amount will constitute 7 billion $, and after that the growth rate will become lower as compared to the previous period (it will decrease up to 8-9%), but nevertheless this dynamics will be characterized by stability (its first order derivative will be a positive number). Thus the profession the key requirements of which are SQL knowledge and skills of DBMS operation, will not only guarantee the employment in future, but will pay on the regular basis.

2. Process of education

Its not enough to communicate useful knowledge, which will be needed for practical purposes; its of paramount importance to do this in the correct way. Surfing on one can come across distance learning sites that have formalized approach to presenting content for the purpose of which investments into such sites have been made, but their efficiency is made null and void due to negligence with which the learning process has been organized. The leader of SQL Exercises project is the high school teacher, who solves such problems within the framework of his job duties. S.I.Moiseenko is an associate professor of the faculty of higher technologies, South Federal University; he is a PhD in physics and mathematics, author of SQL textbook.

3. Community of experts

SQL Exercises, the most popular resource on databases, unfailingly attracts the most experienced professionals in this field, who are present on all the forums of the site and provide aid in finding the optimal solution, considering the task from another perspective or paying attention to some peculiarities of SQL, that are not the matter of general knowledge, or to its concrete case of use. Communication on the forum takes place in the atmosphere of friendliness and well-wishing cooperation; nobody ever blames the newcomer for his ignorance, as it is quite often the case with, which is notorious for its snobbishness and inadequate behavior of its moderators. The ratings of participants are calculated automatically in accordance with the number of the problems solved, the level of complexity of these problems and time of their solving on sql-ex site. Such ratings help the participants to adequately estimate their level of knowledge as well as the level of competence of their interlocutors on the forum.

4. The quality of exercises

All the exercises are grouped into 3 blocks, which can be conventionally called qualification, credit and optimization units. Suffice to say that SQL tasks, set at Microsoft and Oracle certification examinations (as well as those, presented at some other examinations) wont pose any difficulty for the person, who has completed the three blocks of exercises from beginning to end. Basically, on the site there are those exercises that can be characterized as the problems of master class level or, if it can be formulated in this way, Olympiad-class exercises. One can be pretty sure that, the specialist, who has acquired extensive practical experience of solving such problems, wont find SQL tasks, which he will face during his practical work, too difficult to solve.

5. Certification process

One of the major advantages of site is that it provides educational exercises and guidance free of charge. Payment is required only for the certificates of the 1st and the 2nd level of education (in the amount of 40 USD each) (depending on the number of solved problems). When the first certificate is obtained, the second one costs only 20 USD. If the participant chooses to do so, he can make no order for such a certificate the acquired knowledge is his major achievement in any case. However the certificates of win more and more acknowledgement among recruitment agencies and employers who are selecting the personnel for the jobs connected with database operation, as they evidence that the person, awarded with such certificates, is a brilliant specialist in SQL.


As it is often the case, our drawbacks are nothing more than the outcome of our positive traits and vice a versa. I cannot say that I have noticed a lot of negative things about, however there are certain things I would like to point out, and they are immediately connected with the merits of the site.

I have already mentioned the fact that the level of difficulty of the majority of problems presented on site is rather high, they can be even recommended for programming Olympiads. However this can be treated not only as the advantage of the site but also as its drawback. Is it really worthwhile in terms for practical usefulness to solve such problems, if such a high level of difficulty is not really needed for practical purposes? One has an impression that the authors of the problems have not satisfied their Olympiad enthusiasm and there is a certain gap between their methods and reality. There is no denying of the fact that exercise is good for brain, but its better to mark such extremely difficult problems with a special sign and to subdivide them into a separate group. Is it really needed to train people to survive in the jungles of the Amazon river if in all probability no one of them will find himself there ever in his life? It is better to concentrate on more practical things that have more chances to come up in their potential work situations. If someone is ready to find time in order to get acquainted with the exotic tasks of high complexity, he is welcome to have a look at the corresponding section. In order to make such a subdivision it is necessary to verify the level of complexity of the problems. There are certain problems, the level of complexity of which is clearly misinterpreted; the latter is proved by the amount of time spent by the participants on their solving. Initially it was probably assessed in accordance with the considerations of the author, but when certain statistical data is accumulated, the level of complexity should be adjusted in accordance with this data.

Apart from real Olympiad-class problems, the solving of which is really worthwhile (on the condition that one has some time to spend); there are certain problems, the solution of which is artificially complicated because of the limitations set on site. The most essential to my mind is the fact, that the problem should be solved using the single SELECT-query in accordance with the limit of the plan cost. In real life one uses local variables, temporary tables, operational constructions based on conditional operators and cycles, one also programs his own procedure or function using CLR. It is easier and simpler. In addition to that, if in real work situation you solved such a problem using a la method, nobody would be happy about it. On the contrary, people would consider you to be a little bit of an oddball because of the amount of time wasted on it. On the whole, it is clear, where such limitations on site originate from: in the first place measures are taken for no abuser to cause any damage under such settings and in the second place the reason is to safeguard the server against run-away queries. But the limitation prohibiting me to use regardless of the fact that all such problems have been solved and verified on SQL Server 2005 for quite a long time occurs as extremely queer to me.

The number of exercises, the application of SQL to which is unlikely and the presentation of which on the site cannot be sufficiently justified, are quite a separate issue. The most demonstrative example is problem number 59. There is a certain instrument for solution of each problem. For example, hammer is intended to drive in a nail and corkscrew is intended to take the cork out of the bottle, and it is not worthwhile to teach people to do it in any other way. God save them from doing vice a versa at their workplace, they are hardly likely to be praised for it. It is hardly likely that someone of the site visitors is trying to master SQL out of theoretical interest. It is clear, that the majority is going to use the acquired knowledge in practice. SQL is a declarative language, characterized by the functional style of programming and targeted for operation of data sets. It is its advantage. Even taking into account the affection with which the authors treat SQL, it is not worth pretending that it can be used for solving problems of any type, it is not really so. In reality no one would ever go to such extremes, using SQL for solution of algorithmic and imperative problems. If the worst comes to the worst, one can program a function. My conclusion is that this problem is of no practical value. It is of no use and even unfortunate to waste time on such problems, because the artistic skills with which you would cope with their solving in the end are nowhere to be applied in the long run.

Another useless complication is posed by the exercises of the 3-rd database (Ships). In my opinion the main difficulty here is produced by the peculiarities of its scheme, which is characterized by an artificially slipshod design (for example, "the ships, not related to Ships table can be related to Outcomes table), such a wording was purposefully formulated in order to complicate the process of solving, and such a complication has nothing to do with difficulties of real life projects. In real life nobody would ever use such a scheme in work and the first stage of fulfillment of the project would be its normalization and clarification in order to avoid the consequent difficulties when formulating the queries using the slipshod scheme.

Another disadvantage, arising due to the automated mode of fulfillment and check of exercises, is the use of two databases to test a query: the major one and the verification one. If the results of processing of the query have coincided with those of template solution, the problem is considered to be solved. One can scan the results of processing of the query on the major database, but the results for the verification database are not available. You are merely informed on whether the results have coincided or not. The creation of verification database is quite justified, it helps to avoid bare adjustment of the results using the functions select ... union all .Its structure is analogous to that of the major one, but it has its own data set. Such a structure is seemingly logical. But the problem is that the data are distributed much easier in the major database. One has an impression that someone with secret malevolence has stuffed the verification database with extreme examples and exclusive situations that no one would ever think of, considering the structure of data in the major one. The typical type of error is Your query has selected the correct data from the major database, but it has failed verification procedure on the invisible one and it means, that actually the speculative type of SQL study is suggested to the visitor. To my mind this is also an artificial complication, that could occur mostly to academics, having the academic way of thinking, but it has not much in common with real life tasks. In his opinion Dmitry Valuyev ( states in a rather involved way The purpose of the verification database is the safeguard against adjustment of the result; however this is an example of what is called in statistics "general total". Actually it is the neither case. First, the use of verification database does not guarantee that the results wont be adjusted after the process of examining of the structure of data using the method of errors and some indirect information. To adduce such an example I have solved the problem 98 in this way, having got the query at record minimal speed and having spent on its solution the shortest time possible. Second, there is no such a term as general total in real life and the verification database is not an example of it. There are some exercises, upon the fulfillment of which the verification database has been upgraded so to say. The most severe testing procedure, which is applied to the program, wont make you 100% sure that it is free from bugs. It can guarantee only that, within the limits of a given set of initial terms and of some particular distribution of data it will produce the forecasted results. Basically it is the criterion of the correctness of the program. It is a black box, and no other functions are delegated to it. In practice simple economical factors are at play: why should one increase the safety level from 0.999 up to 0.9999, if such works would cost dozens of times more than it has already been spent on the program. Lets consider one more point. In fact all over the world thousands of programmers make thousands queries, that would never be admitted upon the procedure of verification database testing, if there would ever be such a verification database in real life. However the majority of them never undergo the verification database testing process.And one can be 99.9% sure that such a process will never be initiated. Excuse me my impertinence, but I strongly doubt that all the SQL-queries used in applications of South Federal University for example were written in an absolutely and ultimately correct way. But nobody is ever interested in this fact, just because they tend to produce correct results within the concrete data set they are intended to address to. And as long as they do so, nobody will ever learn about it. And if some unforeseen amendment of rare type has been introduced into the data set, and the results have become incorrect, the error is identified and eliminated for that particular case, because the data set can be scanned and it is not necessary to guess what it is like. There is no such a thing as invisible data sets in practice. It is better, it seems to me, to interchange the functions of the major and the invisible database, because using the approach that is applied now (determine the situation in which such a query can fail to work) on site the site should be intended not for developers but for testers.

Print  Print version

Usage of any materials of this site is possible
only under condition of mandatory allocation of the direct link to a site
on each page where used materials are placed.

 Main   Articles    Books Rambler's Top100 Alt   SQL: , ,    SQL Copyright c 2002-2006. All rights reserved.