The SSIS Engine

The SSIS Engine

Before learning about buffers, asynchronous components, and execution trees, you might find it useful to consider this analogy — traffic management. While driving in a big city, have you ever wondered how the traffic system works? It’s remarkable how well coordinated the traffic lights are. In Manhattan, for example, a taxi ride can take you from midtown to downtown in minutes — in part because the lights are timed in a rolling fashion to maintain efficiency. The heavy fine assessed to anyone who “locks the box” (remains in the intersection after the light turns red) demonstrates how detrimental it is to interfere with the synchronization of such a complex traffic grid.

Contrast the efficiency of Manhattan with the gridlock and delays that result from a poorly designed traffic system. We have all been there before — sitting at a red light for minutes despite the absence of traffic on the intersecting streets, and then after the light changes, you find yourself in the same scenario at the next intersection! Even in a light-traffic environment, progress is impeded by poor coordination and inefficient design.

In the case of optimizing traffic patterns in a big city, good traffic management design focuses on easing congestion by keeping traffic moving or redirecting more highly congested areas to less utilized streets. A traffic jam occurs when something slows or blocks the flow of traffic. This causes backpressure behind the cause of the jam as cars queue up waiting to pass.

Bringing this back around to SSIS, in some ways the engine is similar to the grid management of a big city because the SSIS engine coordinates server resources and Data Flow for efficient information processing and management of data backpressure. Part of the process of making package execution efficient requires your involvement. In turn, this requires knowing how the SSIS engine works and some important particulars of components and properties that affect the data processing. That is the purpose of this Understanding and Tuning the Data Flow Engine Topic: to provide the groundwork for understanding SSIS, which will lead to better — that is, more efficient — design.

Understanding the SSIS Data Flow and Control Flow

From an architectural perspective, the difference between the Data Flow and Control Flow is important. One aspect that will help illustrate the distinction is to look at them from the perspective of how the components are handled. In the Control Flow, the task is the smallest unit of work, and a task requires completion (success, failure, or just completion) before subsequent tasks are handled. In the Data Flow, the transformation, source, and destination are the basic components; however, the Data Flow Components function very differently from a task. For example, instead of one transformation necessarily waiting for another transformation to complete before it can proceed with the next set of transformation logic, the components work together to process and manage data.

Comparing Data Flow and Control Flow

Although the Control Flow looks very similar to the Data Flow, with processing objects (tasks and transformations) and connectors that bridge them, there is a world of difference between them. The Control Flow, for example, does not manage or pass data between components; rather, it functions as a task coordinator with isolated units of work. Here are some of the Control Flow concepts:

  • Workflow orchestration
  • Process-oriented
  • Serial or parallel tasks execution
  • Synchronous processing

As highlighted, the Control Flow Tasks can be designed to execute both serially and in parallel — in fact, more often than not there will be aspects of both. A Control Flow Task can branch off into multiple tasks that are performed in parallel or as a single next step that is performed essentially in serial from the first. To show this, Below screen shot illustrates a very simple Control Flow process whose tasks are connected in a linear fashion. The execution of this package shows that the components are serialized — only a single task is executing at a time.

very simple Control Flow process whose tasks are connected in a linear fashion

The Data Flow, conversely, can branch, split, and merge, providing parallel processing, but this concept is different from the Control Flow. Even though there may be a set of connected linear transformations, you cannot necessarily call the process a serial process, because the transformations in most cases will run at the same time, handling subsets of the data in parallel and passing groups of data downstream. Here are some of the unique aspects of the Data Flow:

  • Information-oriented
  • Data correlation and transformation
  • Coordinated processing
  • Streaming in nature
  • Source extraction and destination loading

Similar to the Control Flow shown in Below screen shot, Below screen shot also models a simple Data Flow whose components are connected one after the other. The difference between the Data Flow in Below screen shot and the Control Flow in Below screen shot is that only a single task is executing in the linear flow. In the Data Flow, however, all the transformations are doing work at the same time.

models a ******ebook converter DEMO Watermarks******* simple Data Flow whose components are connected one after the other

In other words, the first batch of data flowing in from the source may be in the final destination step (Currency Rate Destination), while at the same time data is still flowing in from the source.

Multiple components are running at the same time because the Data Flow Transformations are working together in a coordinated streaming fashion, and the data is being transformed in groups (called buffers) as it is passed down from the source to the subsequent transformations.

SSIS Package Execution Time from Package Start to Package Finish

Because a Data Flow is merely a type of Control Flow Task, and more than one Data Flow can be embedded in a package (or none!), the total time it takes to execute a package is measured from the execution of the first Control Flow Task or Tasks through the completion of the last task being executed, regardless of whether the components executing are Data Flow Transformations or Control Flow Tasks. This may sound obvious, but it is worth mentioning because when you are designing a package, maximizing the parallel processing where appropriate (with due regard to your server resources) helps optimize the flow and reduce the overall processing time.

The package shown in below screen shot has several tasks executing a variety of processes and using precedence constraints in a way that demonstrates parallel execution of tasks. The last task, Back Up Database, is the only task that does not execute in parallel because it has to wait for the execution of all the other tasks. Because the Control Flow has been designed with parallelization, the overlap in tasks enables faster package execution than it would if the steps were executed in a serial manner as earlier shown.

The package has several tasks executing a variety of processes and using precedence constraints in a way that demonstrates parallel execution of tasks

Handling Workflows with the Control Flow

Both of the components of the Control Flow have been discussed in SSIS Task, as well as the different types of precedence constraints. Because the Control Flow contains standard workflow concepts that are common to most scheduling and ETL tools, the rest of this Understanding and Tuning the Data Flow Engine Topic focuses on the Data Flow; however, a brief look at Control Flow parallelization and processing is warranted.

The Control Flow, as already mentioned, can be designed to execute tasks in parallel or serially, or a combination of the two. Tasks are also synchronous in nature, meaning a task requires completion before handing off an operation to another process. While it is possible to design a Control Flow containing tasks that are not connected with constraints to other tasks, the tasks are still synchronously tied to the execution of the package. Said in another way, a package cannot kick off the execution of a task and then complete execution while the task is still executing. Rather, the SSIS execution thread for the task is synchronously tied to the task’s execution and will not release until the task completes successfully or fails.

NOTE The synchronous nature of tasks should not be confused with the synchronous and asynchronous nature of transformations in the Data Flow. The concepts are slightly different. In the Data Flow, a transformation’s synchronicity is a matter of communication (how data is passed between transformations), rather than the process orientation of the Control Flow.

Data Processing in the Data Flow

The Data Flow is the core data processing factory of SSIS packages, where the primary data is handled, managed, transformed, integrated, and cleansed. Think of the Data Flow as a pipeline for data. A house, for example, has a primary water source, which is branched to all the different outlets in the house.

When a faucet is turned on, water flows out of it, while at the same time water is coming in from the source. If all the water outlets in a house are turned off, then the pressure backs up to the source to the point where water will no longer flow into the house until the pressure is relieved. Conversely, if all the water outlets in the house are opened at once, then the source pressure may not be able to keep up with the flow of water, and the pressure coming out of the faucets will be weaker. (Of course, don’t try this at home; it may produce other problems!)

The Data Flow is appropriately named because the data equates to the water in the plumbing analogy. The data flows from the Data Sources through the transformations to the Data Destinations. In addition to the flowing concept, there are similarities to the Data Flow pressure within the pipeline. For example, while a Data Source may be able to stream 10,000 rows per second, if a downstream transformation consumes too much server resources, it could apply backward pressure on the source and reduce the number of rows coming from the source. Essentially, this creates a bottleneck that may need to be addressed to optimize the flow. In order to understand and apply design principles in a Data Flow, an in-depth discussion of the Data Flow architecture is merited. Understanding several Data Flow concepts will give you a fuller perspective regarding what is going on under the hood of an executing package. Each of these concepts is addressed over the next few pages:

  • Data buffer architecture
  • Transformation types
  • Transformation communication
  • Execution trees

After reviewing the architecture, you can shift to monitoring packages in order to determine how the Data Flow engine handles data processing.

Memory Buffer Architecture

The Data Flow manages data in groups of data called buffers. A buffer is merely memory that is allocated for the use of storing rows and columns of data where transformations are applied. This means that as data is being extracted from sources into the engine, it is put into these preallocated memory buffers. Buffers are dynamically sized based on row width (the cumulative number of bytes in a row) and other package and server criteria. A buffer, for example, may include 9,000 rows of data with a few columns of data. Below screen shot shows a few groupings of buffers.

few groupings of buffers

Although it is easy to imagine data being passed down from transformation to transformation in the Data Flow, like the flow of water in the pipeline analogy, this is not a complete picture of what is going on behind the scenes. Instead of data being passed down through the transformations, groups of transformations pass over the buffers of data and make in-place changes as defined by the transformations. Think of how much more efficient this process is than copying the data from one transformation to the next every time a transformation specified a change in the data! To be sure, there are times when the buffers are copied and other times when the buffers are held up in cache by transformations. Understanding how and when this happens will enable you to determine the best design to optimize your Data Flow.

This understanding of how memory buffers are managed requires knowing something about the different types of Data Flow Components — transformations, sources, and destinations.

Types of Transformations

The transformations in the Data Flow have certain characteristics that group each into different categories. The base-level differences between them are the way they communicate with each other, and how and when data is handed off from one transformation to another. Evaluating transformations on two fronts provides the background you need to understand how the buffers are managed:

  • Blocking nature: Non-blocking (sometimes called streaming), semiblocking, blocking
  • Communication mechanism: Synchronous and asynchronous

In reality, these classifications are related, but from a practical standpoint, discussing them separately provides some context for data management in the Data Flow.

Non-Blocking, Semi-Blocking, and Blocking Transformations

The most obvious distinction between transformations is their blocking nature. All transformations fall into one of three categories: non-blocking, semi-blocking, or blocking. These terms describe whether data in a transformation is passed downstream in the pipeline immediately, in increments, or after all the data is fully received.

The blocking nature of a transformation is related to what a transformation is designed to accomplish. Because the Data Flow engine just invokes the transformations without knowing what they internally do, there are no properties of the transformation that discretely identify this nature. However, when we look at the communication between transformations in the next section (the synchronous and asynchronous communication), we can identify how the engine will manage transformations one to another.

Non-Blocking, Streaming, and Row-Based Transformations

Most of the SSIS transformations are non-blocking. This means that the transformation logic applied in the transformation does not impede the data from moving on to the next transformation after the transformation logic is applied to the row. Two categories of non-blocking transformations exist: streaming and row-based. The difference is whether the SSIS transformation can use internal information and processes to handle its work or whether the transformation has to call an external process to retrieve information it needs for the work. Some transformations can be categorized as streaming or rowbased depending on their configuration, which are indicated in the list below.

Streaming transformations are usually able to apply transformation logic quickly, using precached data and processing calculations within the row being worked on. In these transformations, it is usually the case that a transformation will not slip behind the rate of the data being fed to it. These transformations focus their resources on the CPUs, which in most cases are not the bottleneck of an ETL system. Therefore, they are classified as streaming. The following transformations stream the data from transformation to transformation in the Data Flow:

  • Audit
  • Cache Transform
  • Character Map
  • Conditional Split
  • Copy Column
  • Data Conversion
  • Derived Column
  • Lookup (with a full-cache setting)
  • Multicast
  • Percent Sampling
  • Row Count
  • Script Component (provided the script is not configured with an asynchronous output, which is discussed in the “Advanced Data Flow Execution Concepts” section)
  • Union All (the Union All acts like a streaming transformation but is actually a semi-blocking transformation because it communicates asynchronously, which is covered in the next section)

The second grouping of non-blocking transformations is identified as rowbased. These transformations are still non-blocking in the sense that the data can flow immediately to the next transformation after the transformation logic is applied to the buffer. The row-based description indicates that the rows flowing through the transformation are acted on one by one with a requirement to interact with an outside process such as a database, file, or component. Given their row-based processes, in most cases these transformations may not be able to keep up with the rate at which the data is fed to them, and the buffers are held up until each row is processed. The following transformations are classified as row-based:

  • DQS Cleansing
  • Export Column
  • Import Column
  • Lookup (with a no-cache or partial-cache setting)
  • OLE DB Command
  • Script Component (where the script interacts with an external component)
  • Slowly Changing Dimension (each row is looked up against the dimension in the database)

Below screen shot shows a Data Flow composed of only streaming transformations. If you look at the row counts in the design UI, you will notice that the transformations are passing rows downstream in the pipeline as soon as the transformation logic is completed. Streaming transformations do not have to wait for other operations in order for the rows being processed to be passed downstream.

Data Flow composed of only streaming transformations

Also, notice in above screen shot that data is inserted into the destination even while transformation logic is still being applied to some of the earlier transformations. This very simple Data Flow is handling a high volume of data with minimal resources, such as memory usage, because of the streaming nature of the Transformation Components used.

Semi-Blocking Transformations

The next category of Transformation Components are the ones that hold up records in the Data Flow for a period of time before allowing the memory buffers to be passed downstream. These are typically called semi-blocking transformations, given their nature. Only a few out-of-the-box transformations are semi-blocking in nature:

  • Data Mining Query
  • Merge
  • Merge Join
  • Pivot
  • Term Lookup
  • Unpivot
  • Union All (also included in the streaming transformations list, but under the covers, the Union All is semi-blocking)

The Merge and Merge Join Transformations are described in detail in The Data Flow Topic and Joining Data Topic but in relation to the semi-blocking nature of these components, note that they require the sources to be sorted on the matching keys of the merge. Both of these transformations function by waiting for key matches from both sides of the merge (or join), and when the matching sorted keys from both sides pass through the transformations, the records can then be sent downstream while the next set of keys is handled. below screen shot shows how a Merge Join within a Data Flow will partially hold up the processing of the rows until the matches are made.

how a Merge Join within a Data Flow will partially hold up the processing of the rows until the matches are made

Typically, the row count upstream of the Merge Join is much higher than the row count just below the Merge Join, because the Merge Join waits for the sorted key matches as they flow in from both sides of the merge. Buffers are being released downstream, just not in a streaming fashion as in the nonblocking Transformation Components. You may also be wondering why there is no Sort Transformation on the right-side source of the Merge Join despite the fact that the transformations require the sources to be sorted. This is because the source data was presorted, and the Source component was configured to recognize that data flowing into the Data Flow was already sorted. Joining Data Topic describes how to set the IsSorted property of a source.

NOTE Semi-blocking transformations require a little more server resources than non-blocking transformations because the buffers need to stay in memory until the right data is received.

In a large data processing situation, the question is how the Merge Join will handle a large data set coming in from one side of the join while waiting for the other set or if one source is slow and the other fast. The risk is exceeding the buffer memory while waiting. However, SSIS 2014 can throttle the sources by limiting the requests from the upstream transformations and sources, thereby preventing SSIS from getting into an out-of-memory situation. This very valuable engine feature of SSIS 2014 resolves the memory tension, enabling a Data Flow to continue without introducing a lot of data paging to disk in the process.

Blocking Transformations

The final category of transformation types is the actual blocking transformation. These components require a complete review of the upstream data before releasing any row downstream to the connected transformations and destinations. The list is also smaller than the list of nonblocking transformations because there are only a few transformations that require “blocking” all the data to complete an operation. Here is the list of the blocking transformations:

  • Aggregate
  • Fuzzy Grouping
  • Fuzzy Lookup
  • Row Sampling
  • Sort
  • Term Extraction
  • Script Component (when configured to receive all rows before sending any downstream)

The two most widely used examples of the blocking transformations are the Sort and the Aggregate; each of these requires the entire data set before handing off the data to the next transformation. For example, in order to have an accurate average, all the records need to be held up by the Aggregate Transformation. Similarly, to sort data in a flow, all the data needs to be available to the Sort Transformation before the component knows the order in which to release records downstream. Below screen shot shows a Data Flow that contains an Aggregate Transformation. The screen capture of this process shows that the entire source has already been brought into the Data Flow, but no rows have been released downstream while the transformation is determining the order.

Data Flow that contains an Aggregate Transformation

With a Blocking Component in the Data Flow (refer to above screen shot), the data is no longer streaming through the Data Flow. Instead, data is held up so that all the rows remain in the blocking transformation until the last row flows through the transformation.

NOTE Blocking transformations are usually more resource intensive than other types of transformations for two main reasons. First, because all the data is being held up, either the server must use a lot of memory to store the data or, if the server does not have enough memory, a process of file staging happens, which requires the I/O overhead of staging the data to disk temporarily. Second, these transformations usually put a heavy burden on the processor to perform the work of data aggregation, sorting, or fuzzy matching.

Synchronous and Asynchronous Transformation Outputs

Another important differentiation between transformations is related to how transformations that are connected to one another by a path communicate. While closely related to the discussion on the blocking nature of transformations, synchronous and asynchronous refer more to the relationship between the Input and Output Component connections and buffers.

Some transformations have an Advanced Editor window, which, among other things, drills into specific column-level properties of the transformations’ input and output columns, which is useful in understanding the difference between synchronous and asynchronous outputs. Below screen shot shows the Advanced Editor of the Sort Transformation, highlighting the Input and Output Properties tab. This particular transformation has a Sort Input and Sort Output group with a set of columns associated with each.

Advanced Editor of the Sort Transformation, highlighting the Input and Output Properties tab

When a column is highlighted, the advanced properties of that column are displayed on the right, as shown in the figure. The advanced properties include such things as the data type of the column, the description, and so on. One important property to note is the LineageID. This is the integer pointer to the column within the buffers. Every column used in the Data Flow has at least one LineageID in the Data Flow. A column can have more than one LineageID as it passes through the Data Flow based on the types of transformation outputs (synchronous or asynchronous) that a column goes through in the Data Flow.

Asynchronous Transformation Outputs

It is easier to begin with the asynchronous definition because it leads into a comparison of the two kinds of transformation outputs, synchronous and asynchronous. A transformation output is asynchronous if the buffers used in the input are different from the buffers used in the output. In other words, many of the transformations cannot both perform the specified operation and preserve the buffers (the number of rows or the order of the rows), so a copy of the data must be made to accomplish the desired effect.

The Aggregate Transformation, for example, may output only a fraction of the number of rows coming into it; or when the Merge Join Transformation has to join two data sets together, the resulting number of rows may not be equivalent to the number of input rows. In both cases, the buffers are received, the processing is handled, and new buffers are created. For example, returning to the Advanced Editor of the Sort dialog shown in

Above screen shot, note the LineageID property of the input column. In this transformation, all the input columns are duplicated in the output columns list. In fact, as below screen shot shows, the output column highlighted for the same input has a different LineageID.

output column highlighted for the same input has a different LineageID.

The LineageIDs are different for the same column because the Sort Transformation output is asynchronous, and the data buffers in the input are not the same buffers in the output; therefore, a new column identifier is needed for the output. In the preceding example, the input LineageID is 181, whereas in the output column the LineageID is 150.

A list doesn’t need to be included here, because all the semi-blocking and blocking transformations already listed have asynchronous outputs by definition — none of them can pass input buffers on downstream because the data is held up for processing and reorganized.

NOTE One of the SSIS engine components is called the buffer manager. For asynchronous component outputs, the buffer manager is busy at work, decommissioning buffers for use elsewhere (in sources or other asynchronous outputs) and reassigning new buffers to the data coming out of the transformation. The buffer manager also schedules processor threads to components as threads are needed.

Synchronous Transformation Outputs

A synchronous transformation is one in which the buffers are immediately handed off to the next downstream transformation at the completion of the transformation logic. If this sounds like the definition for streaming transformations, that’s because there is almost complete overlap between streaming transformations and synchronous transformations. The word buffers was intentionally used in the definition of synchronous outputs, because the important point is that the same buffers received by the transformation input are passed out the output.

The LineageIDs of the columns remain the same as the data is passed through the synchronous output, without a need to copy the buffer data and assign a new LineageID (as discussed previously for asynchronous transformation output).

In this example, each transformation, including the Conditional Split, has asynchronous outputs. This means that there are no buffer copies while this Data Flow runs. For the Source component called Sales, several of the columns are used in both the destinations. Now consider Below screen shot, which shows the Advanced Editor of the source and highlights the TaxAmt column, which has a LineageID of 318.

the Advanced Editor of the source and highlights the TaxAmt column, which has a LineageID of 318.

This LineageID for TaxAmt (as well as the other columns) is preserved throughout the Data Flow all the way to the destination. Below screen shot shows the Advanced Editor for one of the destination components, Reseller Sales, and indicates that the LineageID of 318 has been preserved through the entire Data Flow.

the Advanced Editor for one of the destination components

In fact, if you look at the LineageID of TaxAmt in the Internet Sales destination, you would also find that it is 299. Even for the Conditional Split (because it has asynchronous outputs), the data is not copied to new buffers; rather, the Data Flow engine maintains pointers to the right rows in the buffers, so if there is more than one destination, the data can be inserted into each destination without requiring data to be copied. This explains why the execution of the Data Flow shown earlier in below screen shot allows the data to flow in a streaming way through the Data Flow.

NOTE A transformation is not limited to a single synchronous output. Both the Multicast and the Conditional Split can have multiple outputs, but all the outputs are synchronous.

With the exception of the Union All, all the non-blocking transformations listed in the previous section also have synchronous outputs. The Union All, while it functions like a streaming transformation, is really an asynchronous transformation. Given the complexity of unioning multiple sources together and keeping track of all the pointers to the right data from the different source inputs, the Union All instead copies the upstream data to new buffers as it receives them and passes the new buffers off to the downstream transformations.

NOTE Synchronous transformation outputs preserve the sort order of incoming data, whereas some of the asynchronous transformations do not. The Sort, Merge, and Merge Join asynchronous components, of course, have sorted outputs because of their nature, but the Union All, for example, does not.

Source and Destination Components

Source and Destination components are integral to the Data Flow and therefore merit brief consideration in this Understanding and Tuning the Data Flow Engine Topic. In fact, because of their differences in functionality, sources and destinations are therefore classified differently.

First of all, in looking at the advanced properties of a Source component, the source will have the same list of external columns and output columns. The external columns come directly from the source and are copied into the Data Flow buffers and subsequently assigned LineageIDs. While the external source columns do not have LineageIDs, the process is effectively the same as an asynchronous component output. Source components require the allocation of buffers, where the incoming data can be grouped and managed for the downstream transformations to perform work against. 

Destination components, conversely, de-allocate the buffer data when it is loaded into the destinations. The advanced properties of the Destination component include an External Column list, which represents the destination columns used in the load. The input columns are mapped to this External Column list on the Mapping page of the Destination component editor. In the advanced properties, you should note that there is no primary Output Container (besides the Error Output) for the Destination component, as the buffers do not flow through the component but rather are committed to a Destination component as a final step in the Data Flow.

Advanced Data Flow Execution Concepts

The preceding discussion of transformation types and how outputs handle buffers leads into a more advanced discussion of how SSIS coordinates and manages Data Flow processing overall. This section ties together the discussion of synchronous and asynchronous transformations to provide the bigger picture of a package’s execution.

Relevant to this discussion is a more detailed understanding of buffer management within an executing package based on how the package is designed.

Execution Trees

In one sense, you have already looked at execution trees, although they weren’t explicitly referred to by this name. An execution tree is a logical grouping of Data Flow Components (transformations, sources, and destinations) based on their synchronous relationship to one another. Groupings are delineated by asynchronous component outputs that indicate the completion of one execution tree and the start of the next.

Below screen shot shows the execution trees of a moderately complex Data Flow that uses multiple components with asynchronous outputs. The “paths” indicated are numbered based on the SSIS Data Flow logging, not the order in which the data flows. See the PipelineExecutionTrees log example in the “Monitoring Data Flow Execution” section of this Understanding and Tuning the Data Flow Engine Topic.

the execution trees of a moderately complex Data Flow

Recall that components with asynchronous outputs use different input buffers. The input participates in the upstream execution tree, while the asynchronous output begins the next execution tree. In light of this, the execution trees for above screen shot start at the Source components and are then completed, and a new execution tree begins at every asynchronous transformation.

Execution trees are base 0, meaning you count them starting with a 0. In the next section, you will see how the pipeline logging identifies them. Although the execution trees seem out of order, you have used the explicit order given by the pipeline logging.

In the next section, you will look at ways to log and track the execution trees within a Data Flow, but for now the discussion focuses on a few principles that clarify what happens in an execution tree.

When SSIS executes a package, the buffer manager defines different buffer profiles based on the execution trees within a package. All the buffers used for a particular execution tree are identical in definition. When defining the buffer profile for each execution tree, the SSIS buffer manager looks at all the transformations used in the execution tree and includes every column in the buffer that is needed at any point within the execution tree. Note that execution tree path #5 in above screen shot contains a Source component, a Derived Column Transformation, and a Lookup. Without looking at the source properties, the following list defines the four columns that the Source component is using from the source:

  • CurrencyCode
  • CurrencyRate
  • AverageRate
  • EndofDayRate

The Derived Column Transformation adds two more columns to the Data Flow: Average_Sale and Audit_Date. Finally, the Lookup Transformation adds another three columns to the Data Flow.

Added together, the columns used in these three components total nine. This means that the buffers used in this execution tree will have nine columns allocated, even though some of the columns are not used in the initial components. Optimizing a Data Flow can be compared to optimizing a relational table, where the smaller the width and number of columns, the more that can fit into a Data Flow buffer. This has some performance implications, and the next section looks in more detail at optimizing buffers.

When a buffer is used in an execution tree and reaches the transformation input of the asynchronous component (the last step in the execution tree), the data is subsequently not needed because it has been passed off to a new execution tree and a new set of buffers. At this point, the buffer manager can use the allocated buffer for other purposes in the Data Flow.

NOTE One final note about execution trees — the process thread scheduler can assign more than one thread to a single execution tree if threads are available and the execution tree requires intense processor utilization. Each transformation can receive a single thread, so if an execution tree has only two components that participate, then the execution tree can have a maximum of two threads. In addition, each source component receives a separate thread.

One advanced property of the Data Flow is the EngineThreads property. In the Control Flow, when a Data Flow Task is highlighted, this property appears in the Properties window list (see below screen shot).

It is important to modify the EngineThreads property of the Data Flow so that the execution trees are not sharing process threads, and extra threads are available for large or complex execution trees. Furthermore, all the execution trees in a package share the number of processor threads allocated in the EngineThreads property of the Data Flow. A single thread or multiple threads are assigned to an execution tree based on availability of threads and complexity of the execution tree.

In the last section of this Understanding and Tuning the Data Flow Engine Topic, you will see how the threads available in a Data Flow are allocated to the execution trees. The value for EngineThreads does not include the threads allocated for the number of sources in a Data Flow, which are automatically allocated separate threads.

Monitoring Data Flow Execution

Built into SSIS 2014 is the capability to report on Data Flow performance and identify execution trees and threads. The reports and log events can be very useful in understanding your Data Flow and how the engine is managing buffers and execution.

First of all, pipeline logging events are available in the Logging features of SSIS. As shown in the following list, several logging events provide details about what is happening in the Data Flow. Tutorial Online describes each of these in detail, with examples, at http://msdn.microsoft.com/en-us/ library/ms141122(v=SQL.120).aspx.

  • OnPipelinePostEndOfRowset
  • OnPipelinePostPrimeOutput
  • OnPipelinePreEndOfRowset
  • OnPipelinePrePrimeOutput
  • OnPipelineRowsSent
  • PipelineBufferLeak
  • PipelineComponentTime
  • PipelineExecutionPlan
  • PipelineExecutionTrees
  • PipelineInitialization

NOTE When executing a package on the SSIS server, you can turn on the Verbose logging to capture all the events. This is found on the Advanced tab of the Execute Package dialog under Logging level.

To better understand execution trees, you can capture the PipelineExecutionTrees log event. To do so, create a new log entry through the logging designer window under the SSIS menu’s Logging option. The pipeline events are available only when a Data Flow is selected in the tree menu navigator of the package executable navigator, as shown in below screen shot.

pipeline events are available only when a Data Flow is selected in the tree menu navigator of the package executable navigator

On the Details tab of this Configure SSIS Logs dialog, the two execution information log events just listed are available to capture. When the package is run, these events can be tracked to the selected log provider as defined. However, during development, it is useful to see these events when testing and designing a package. SSIS includes a way to see these events in the SQL Server Data Tools as a separate window. The Log Events window can be pulled up either from the SSIS menu by selecting Log Events or through the View menu, listed under the Other Windows submenu.

When the package is executed in design time through the interface, the log events selected are displayed in the Log Events window. For each Data Flow, there is one event returned for the PipelineExecutionPlan event and one for the PipelineExecutionTrees event, as shown in below screen shot. 

Note that all pipeline events selected in the Logging configuration are included in the Log window. To capture the details for a more readable view of the Message column, simply right-click the log entry and copy, which puts the event message onto the clipboard.

Pipeline Execution Trees Log Details

The PipelineExecutionTrees log event describes the grouping of transformation inputs and outputs that participate in each execution tree. Each execution tree is numbered for readability. The following text comes from the message column of the PipelineExecutionTrees log entry from the execution of the Data Flow shown in below screen shot:

Begin Path 0
Customer Validation.Outputs[OLE DB Source Output]; Customer
Filter Valid Customers.Inputs[Merge Join Right Input]; Filter
Valid Customers
End Path 0
Begin Path 1
Filter Valid Customers.Outputs[Merge Join Output]; Filter Valid
Union All.Inputs[Union All Input 1]; Union All
End Path 1
Begin Path 2
Prior Weeks.Outputs[OLE DB Source Output]; Prior Weeks
Union All.Inputs[Union All Input 2]; Union All
End Path 2
Begin Path 3
Sort.Outputs[Sort Output]; Sort
Filter Valid Customers.Inputs[Merge Join Left Input]; Filter Valid
End Path 3
Begin Path 4
Union All.Outputs[Union All Output 1]; Union All
Monthly Vendor Details.Inputs[SQL Server Destination Input];
Monthly Vendor
End Path 4
Begin Path 5
Weekly Vendor Export.Outputs[Flat File Source Output]; Weekly
Vendor Export
Average Calc.Inputs[Derived Column Input]; Average Calc
Average Calc.Outputs[Derived Column Output]; Average Calc
Customer Attributes.Inputs[Lookup Input]; Customer Attributes
Customer Attributes.Outputs[Lookup Match Output]; Customer
Sort.Inputs[Sort Input]; Sort
End Path 5

In the log output, each execution tree evaluated by the engine is listed with a begin path and an end path, with the transformation inputs and outputs that participate in the execution tree. Some execution trees may have several synchronous component outputs participating in the grouping, while others may be composed of only an input and output between two asynchronous components. As mentioned earlier, the execution trees use base 0, so the total number of execution trees for your Data Flow will be the numeral of the last execution tree plus one. In this example, there are six execution trees. A quick way to identify synchronous and asynchronous transformation outputs in your Data Flow is to review this log. Any transformation for which both the inputs and the outputs are contained within one execution tree is synchronous. Conversely, any transformation for which one or more inputs are separated from the outputs in different execution trees has asynchronous outputs

Pipeline Execution Reporting

SSIS 2014 comes with a set of execution and status reports that highlight the current executions, history, and errors of package execution. Administering SSIS Topic covers the reports in general. These reports are available only if you have your packages in the project deployment model, have deployed them to an SSIS server, and run the package on the server. In addition to the report highlighted later, you can also query the execution data in the SSISDB database on the server. The log information named in SSIS 2014 is available in the catalog. In addition, there is a table-value function catalog.dm_execution_performance_counters that can provide execution instance-specific information such as the number of buffers used.

One very valuable data collection point for Data Flow execution is the “Active Time” (in seconds) of each Data Flow Component. If you want to know how long a transformation is being used during a Data Flow execution, you can compare the active time to the total time of the Data Flow. In order to capture these details, you need to enable the performance logging level when the package is executed. Figure 16-16 shows the Advanced tab of the Execute Package window when executing a package on the SSIS server.

ssis engine

This logging level is also available in SQL Server Agent when you schedule a package to be run through the SQL Server Integration Services Package job step.

The benefit of logging this level of detail for a package is that it retains the performance details of the Data Flow. If you run the Execution Performance report (right-click the package in the SSIS Catalog project and run the All Executions report, then drill down to Execution Performance), you can see the history of the package, as well as the Data Flow Components and how long each were active. below screen shot shows a sample Execution Performance report.

sample Execution Performance report

As you can imagine, if you are having performance problems with your Data Flows, you can run this report and quickly identify the trouble spots that you need to work on. The next section describes ways to efficiently design, optimize, and tune your Data Flow.

0 Responses on The SSIS Engine"

Leave a Message

Your email address will not be published. Required fields are marked *

Copy Rights Reserved © Mindmajix.com All rights reserved. Disclaimer.