We have a few Software Design Engineer Test openings in the data replication and synchronization team of SQL Server, Microsoft Corporation. Prospective candidates are encouraged to contact li.zhang@microsoft.com. The following are job descriptions.

 Description 1:

Do you get excited about distributed systems and the challenges this domain presents? Do you want to work on deep and challenging technical problems that have real world applicability and long-term industry-wide implications? Would you like to be a part of the team building the next generation of replication and synchronization technologies for Microsoft? If you are excited about tackling problems in this space we want to speak with you. The Data Replication and Synchronization team in SQL Server has been tasked with delivering innovative synchronization and replication techniques that are core building blocks for data movement within an organization and/or across the Internet. Solutions will span from high-end enterprise customers to the disconnected mobile work force.

The Data Replication and Synchronization test team is seeking highly qualified SDETs to work on replication improvements of the next release of SQL Server. The team is very dynamic with ample opportunities for personal career development. You will participate in product architecture and design reviews, analyze the customer requirements, develop feature test plans, work with developers and program managers to ensure proper test coverage, design and implement automated tests in C++, C# and/or transact-SQL, debug problems, develop solid infrastructure and new tools. Qualifications include solid understanding of SQL Server or other database management systems, three years experience of testing commercial software products, the ability to multitask, and good communication skills. Prior experience and knowledge of replication technology is preferred. Proficiency of programming in one or more high level languages, such as C, C++ or C#, is required. Proficiency in transact-SQL programming is a plus. A bachelor’s degree in Computer Science or related technical field is preferred.

Description 2:

The Data Replication and Sync team is working to make sure that you can have your important data available no matter what PC, Device, or Web Page you are using. Have you ever considered that as computer technology becomes more pervasive, users are going to be faced more and more with the problem of getting access to something that they have on another machine or device? Data synchronization is going to be the cornerstone technology for solving this problem. By having assets on the PC, the server, the cloud, and on devices, Microsoft is in a better position than any other company in the industry to provide a solution that will allow you to move your data wherever you need it. In the Data Replication and Sync team we are working on a project called Harmonica to do exactly that.

Do you want to work in a startup environment? We’re moving very fast with short milestones to enabling us to align with customers and partners. In less than a year we’ve already deployed this technology to the Windows Live team that provided the end to end solution for Favorites synchronization. Do you want to work with cool groups? We’re working with teams in Windows Live, Zune, and Office just to name a few. Want to work on cool technology? We are driving the core metadata layer for data synchronization that enables sharing in a multi-master mesh fashion and delivering the toolkit that will allow other teams to use this metadata to solve their synchronization problems. It is inherently distributed. It is algorithmically focused. We have lots of hard problems that still need to be solved and we need smart passionate people to work on them.

A successful candidate will have:

· Excellent coding skills in C, C++, or C#

· Great ability to work with other teams and people

· Excellent problem solving skills

· A proven track record for shipping quality software on time

· The ability to work within product teams and across product groups

· Passion for engineering excellence and learning

· Three or more years of experiencing in testing commercial software product

· A BS or MS in computer science or related field

Merge replication filters may affect replication performance significantly, if the filters are not designed properly. I have a script that will check the following issues in merge filters:

• Checking existence of subset filter and join filters on the same article.
• Checking existence of multiple join filters on the same article
• Checking existence of circular join filters in articles
• Finding deepest join filter chains of articles
• Checking reference to un-published tables in article filters

You can run it in a database enabled for merge replication. All rights reserved.

set nocount on

if DATABASEPROPERTYEX(DB_NAME(), 'IsMergePublished') = 0
  begin
    print 'Current database is not published for merge replication.'
    goto done
  end

print ' '
print '*****************************************************************************************************'
print '* Checking existence of subset filter and join filters on the same article ...'
print '*****************************************************************************************************'
print ' '

select 'pubname' = p.name, 'artname' = a.name
into #multiplefilters1
from sysmergearticles a, sysmergesubsetfilters f, sysmergepublications p
where a.artid = f.artid
and a.pubid = f.pubid
and a.subset_filterclause is not null
and a.subset_filterclause <> ''
and a.pubid = p.pubid

if exists (select * from #multiplefilters1)
  begin
    print 'The following articles have both subset filter and join filters. The logical relationship of these '
    print 'filters is OR and will be implemented as UNION in the article view. This may have performance impact.'
    print 'You may wish to consider re-implementing them.'
    print ' '
    select 'Publication Name' = pubname, 'Article Name' = artname
      from #multiplefilters1
    print ' '
  end

drop table #multiplefilters1

print '*****************************************************************************************************'
print '* Checking existence of multiple join filters on the same article ...'
print '*****************************************************************************************************'
print ' '

select f.pubid, f.artid
into #multiplefilters2
from sysmergesubsetfilters f
group by f.pubid, f.artid
having count(*) > 1

if exists (select * from #multiplefilters2)
  begin
    print 'The following articles have multiple join filters. The logical relationship of these filters is OR '
    print 'and will be implemented as UNION in the article view. This may have performance impact. '
    print ' '
    select 'Publication Name' = p.name, 'Article Name' = a.name, 'Join Article Name' = f.join_articlename, 'Filter Name' = f.filtername
      from #multiplefilters2 m, sysmergearticles a, sysmergesubsetfilters f, sysmergepublications p
     where m.artid = a.artid
       and m.pubid = a.pubid
       and m.artid = f.artid
       and m.pubid = f.pubid
       and p.pubid = m.pubid
     order by 1, 2
    print ' '
  end

drop table #multiplefilters2

print '*****************************************************************************************************'
print '* Checking existence of circular join filters in articles ...'
print '*****************************************************************************************************'
print ' '

select distinct 'pubname' = p.name, 'artname' = a.name, f.join_articlename, 'level' = NULL
into #multiplefilters3
from sysmergesubsetfilters f, sysmergepublications p, sysmergearticles a
where f.pubid = a.pubid
  and f.artid = a.artid
  and f.pubid = p.pubid

declare @level int
select @level = 0
while @level = 0 OR @@rowcount > 0
  begin
    select @level = @level + 1
    update m3a
       set m3a.level = @level
      from #multiplefilters3 m3a
     where m3a.level is NULL
       and m3a.join_articlename not in (select artname from #multiplefilters3 m3b
     where m3b.level is NULL and m3a.pubname = m3b.pubname)
  end

select @level = 0
while @level = 0 OR @@rowcount > 0
  begin
    select @level = @level - 1
    update m3a
       set m3a.level = @level
      from #multiplefilters3 m3a
     where m3a.level is NULL
       and m3a.artname not in (select join_articlename from #multiplefilters3 m3b
     where m3b.level is NULL and m3a.pubname = m3b.pubname)
  end

if exists (select * from #multiplefilters3 where level is NULL)
  begin
    print 'The following articles have circular join filters with other articles. This may have performance impact. '
    print ' '
    select 'Publication Name' = pubname, 'Article Name' = artname, 'Join Article Name' = join_articlename
      from #multiplefilters3
     where level is NULL
     order by 1, 2, 3
    print ' '
  end

print '*****************************************************************************************************'
print '* Finding deepest join filter chains of articles ...'
print '*****************************************************************************************************'
print ' '

declare @max int, @min int, @start int, @end int, @pubname sysname, @artname sysname, @sequence int
select @max = max(level), @min = min(level) from #multiplefilters3
where level is not null
if @max >= 0 - @min
  select @start = @max, @end = 1
else
  select @start = -1, @end = @min

declare #cursor_deepest_articles cursor for
select pubname, artname from #multiplefilters3 where level = @start

print 'The following artile sets have the deepest level join filters in the database. '
print ' '

select @sequence = 1
open #cursor_deepest_articles
fetch #cursor_deepest_articles into @pubname, @artname
while @@fetch_status <> -1
  begin
    create table #articlechain (id int identity primary key, pubname sysname, artname sysname)
    insert #articlechain (pubname, artname) values (@pubname, @artname)
    if @max >= 0 - @min
      select @start = @max, @end = 1
    else
      select @start = -1, @end = @min

    while @start >= @end
      begin
        select @artname = join_articlename
          from #multiplefilters3
         where level = @start
           and artname = @artname
           and pubname = @pubname
        insert #articlechain (pubname, artname) values (@pubname, @artname)
        select @start = @start - 1
      end
    print 'Article Set #' + convert(varchar(32), @sequence)
    print ' '
    select 'Publication Name' = pubname, 'Article Name ' = artname from #articlechain order by id desc
    print ' '
    drop table #articlechain
    fetch #cursor_deepest_articles into @pubname, @artname
    select @sequence = @sequence + 1
  end
deallocate #cursor_deepest_articles

drop table #multiplefilters3

print '*****************************************************************************************************'
print '* Checking reference to un-published tables in article filters ...'
print '*****************************************************************************************************'
print ' '

declare #csr_filtered_articles cursor for
select a.pubid, a.artid, a.sync_objid
from sysmergearticles a
where a.subset_filterclause <> ''
and a.subset_filterclause is not null
or exists
(select * from sysmergesubsetfilters f
where a.artid = f.artid and a.pubid = f.pubid)

create table #filter_article_dependencies (objid int primary key, expanded bit null)

declare @pubid uniqueidentifier, @artid uniqueidentifier, @sync_objid int

open #csr_filtered_articles
fetch #csr_filtered_articles into @pubid, @artid, @sync_objid
while @@fetch_status <> -1
  begin
    truncate table #filter_article_dependencies
    insert #filter_article_dependencies (objid, expanded) values (@sync_objid, NULL)
    while exists (select * from #filter_article_dependencies where expanded IS NULL)
      begin
        insert #filter_article_dependencies (objid, expanded)
        select distinct d.depid, 0
          from sysdepends d, #filter_article_dependencies f
         where d.id = f.objid
           and f.expanded IS NULL
           and not exists (select * from #filter_article_dependencies f2
                            where f2.objid = d.depid)
        update #filter_article_dependencies set expanded = 1 where expanded IS NULL
        update #filter_article_dependencies set expanded = NULL where expanded = 0
      end
    select f.objid
      into #nonpublishedtables
      from #filter_article_dependencies f, sysobjects o
     where f.objid = o.id
       and o.xtype = 'U'
       and f.objid not in (select objid from sysmergearticles where pubid = @pubid)

    if exists (select * from #nonpublishedtables)
      begin
        select @pubname = name from sysmergepublications where pubid = @pubid
        select @artname = name from sysmergearticles where pubid = @pubid and artid = @artid
        print ' '
        raiserror('Publication [%s] article [%s] references in its subset or join filter(s) ', 0, -1, @pubname, @artname)
        raiserror('the following tables that are not published. This may cause unexpected results at the subscriber.', 0, -1)
        select 'Table name' = object_name(objid) from #nonpublishedtables
      end
    drop table #nonpublishedtables
    fetch #csr_filtered_articles into @pubid, @artid, @sync_objid
  end
deallocate #csr_filtered_articles
drop table #filter_article_dependencies

done: