What needs improvement with SnapLogic?

I think SnapLogic can be improved by implementing more seamless logging of information. Currently, this is done through an error, but it would be best to have logging available on each snap that you can turn on and off, allowing you to configure a logging mechanism for where it logs. By having each snap available for logging, you can log all the information instead of creating a separate pipeline for logging. Additionally, I suggest caching for snaps that are recursively called, particularly when logging in, authenticating, doing actions, logging out, and then repeating the process, which would significantly speed up operations, especially when querying an Active Directory because retrieving many members in Java takes about five seconds, whereas in SnapLogic, it takes a minute and a half for the same operation. Regarding the new dedupe snap, I think the default setting should be changed from the current 80% match to an exact match, allowing for changes afterward, as I ran into issues during regression testing where a minor difference led to duplicates being flagged unintentionally.

I see very few areas where SnapLogic needs improvement. There are some very fringe cases we have had to abandon due to minor complexity and time pressures, such as the ability to read a PDF and isolate tagged data areas on that PDF to extract data, which we did not find very easy to start with. However, we had the limitation of needing to turn this around in a matter of days, so we did not have time to investigate further. Other than that, I find it very difficult to identify areas for improvement; I think it is pretty industry-leading as far as these tools go. Comparing it to other products, I believe the return on investment has been met within a year and then some, whereas previously we wasted a lot of money trying to deliver the same things. Therefore, I have no real notes at this point, but I think consistent evolution is essential, and I look forward to seeing what they do next. Support has been exemplary.

One thing that comes to mind about SnapLogic is that the product is always improving, which is a great thing, but the changes in the UI and the way things keep moving around mean that a level of stability would help achieve a higher adoption ratio. The changes in the UI have frustrated a few people, which is why a nine for SnapLogic is justified. Additionally, some of the snaps could do with greater functionality. For example, the SOAP snap could be opened up to allow broader usage. Regarding SnapLogic's AI capabilities and its governance and security, as you determine the model that it uses, this is very much on the developer or implementer to provide that level of control. SnapLogic is open to lots of models and works with any AI model out there, and that exposes a risk. This creates a risk that you could provide your own sensitive data out to a public model. Therefore, it is very much on the developer to make sure that security and the level of data retention is kept in-house. Regarding SnapLogic's AI capabilities, the accuracy and reliability of output is subject to hallucinations, just any AI model. The accuracy is no better or worse than any other platform. In some ways, it does help with the prompting and helps make a better prompt, which will give back better data, but it is no better or worse than ChatGPT.

From our development perspective, we have requested SnapLogic to enhance their SnapGPT to be friendlier, improve documentation capabilities, and enable more Cursor-based type development. While not all developers are fond of the current environment, SnapLogic has been responsive to our requests for improvement, often providing fast solutions for testing and implementation. Additionally, we wished for a better implementation of the GitHub integration, as the existing method resembles older practices, where pushing a pull request involves pushing an entire package. Fortunately, they have implemented new features that allow us to be more flexible in this regard, but we still desire the ability to manage multiple folders more efficiently rather than having to create a separate repository for each folder.

Areas for improvement for SnapLogic include enhancing the documentation, especially by providing more complex examples in each Snappack, to cover various use cases that different team members may encounter. Another improvement I suggest is enhancing the monitor page on the dashboard to allow quicker and more efficient access to historical data, as it currently becomes slow and sometimes hangs when accessing older data.

More native support, particularly around the Agent Creator AI agent tool, would be beneficial. Expanding to have native Anthropic support would be beneficial for our organization specifically. More native support for other large-scale LLMs such as Anthropic would be helpful. Additionally, more training to address the learning curve would be valuable. However, there is a good amount of information in the knowledge base, and Snap GPT has been extremely beneficial.

SnapLogic can be improved. My only initial suggestion is that the AI agent is a little overly friendly. That is the main thing for now regarding needed improvements. There is some confusion in the documentation, but the overall product seems very good, and those are the additional improvements I would like to see in SnapLogic.

I think SnapLogic could be improved if the Snaplex has a way to change for people that want to write code, or to convert code to the Snaplex. There should be a way of switching the Snaplex into a base code so that you can add to the code and then save it, which would then be converted into logic on the front end. Having an alternative code injection into each Snaplex would be beneficial. For instance, in the Dynamics search Snaplex, sometimes you might want to search and filter by other entities. Right now you need to connect other entities together to accomplish that, but this should be possible if you have a way of injecting code and making SnapLogic run through the code and calling both entities together and doing the filter and pumping out results to be used, rather than each Snaplex doing only one thing.

SnapLogic currently offers SnapGPT with more capabilities compared to earlier versions, but there is still room for improvement compared to other AI tools, particularly in areas such as viewing historical data and automatic file reading which would be very useful. The interface still has options for improvement. While it has been enhanced in recent versions, it is often difficult to identify snaps when developing large pipelines, requiring a lot of zooming. A simpler option to see the snaps would be beneficial. After execution, it is challenging to know how much data has been processed, and there is a need for better AI integration since it lags behind other tools.

I think that the biggest improvement needed for SnapLogic would be the learning curve. Without the use of the GPT, I think we would have really struggled as a business to adopt it. I think it is essential to provide more training opportunities that are offered as part of the package that comes with buying the tool and having the contract. When we asked about having learning materials and training materials, it came at an extra expense on top of the existing price for the tool. In reality, if you had a more mainstream tool such as a Databricks or a Microsoft tool, that documentation and training would be public for you to access because there would be many things on Reddit or other platforms that people have posted. I know that Microsoft does offer free training material. I think there would be wider availability of training resources that do not rely on the GPT to support it as part of your package effectively.

SnapLogic could improve its notification system. When a pipeline fails, the notification sometimes comes back as not very useful. I had a recent case where it just told me that a snap had failed unexpectedly without providing any reason, which is not helpful for troubleshooting, especially on the live system. To resolve this, I had to manually validate the pipeline to find the issue, which allowed me to see the data running into those individual snaps and what the error was. I would like to see if the execution can show the data flowing into those individual snaps and provide the same kind of error that I receive when validating, as that would be much more useful. I would give SnapLogic a higher rating if the error handling and notification were more specific and user-friendly to read. Apart from that, I do not think there is anything else that I can identify.

One thing I do not like about SnapLogic is some of the pre-built snap packs. If I want to integrate with a new system, I have to pay to get access to those. There have been a couple of instances of just different account types that I was not able to get working within SnapLogic and had to use a third party to get that data. Generally having to pay more to get access to the full feature set is a downside. Regarding SnapLogic's AI capabilities, I find that the AI capabilities for the most part are locked behind pay, and I have not purchased many of them. I do use SnapGPT occasionally and know some other teammates who have recently gotten access and have not used the platform before tried that, but as we reviewed the pipelines that it built, they were not exactly great and we had to go through and fix a lot of things for them. It is nice when I am developing and trying to figure out an expression, as it is generally quite useful, but I cannot really speak much to the security and governance portion of it. Regarding SnapLogic's AI capabilities, I think its accuracy and reliability of output are quite good with the expressions, but in my limited experience, the pipeline builder portions of it have not been great.

SnapLogic definitely needs some kinds of improvements as per the current market and current AI world and current technology. At this moment, I can say there is one feedback on how SnapLogic can improve. There is a very small example: in SnapLogic, if we have to whitelist so many IPs, then you have to go one by one and add IPs manually. There should be some kind of a mechanism so that it can take input from any CSV files or something so that it can do bulk uploads. The IP whitelisting is the area that SnapLogic definitely has to improve. It is a bit painful because if you want to whitelist some Salesforce IPs, Salesforce provides complete IP ranges that you have to go through manually. Sometimes if you mistakenly select something, then all your existing IPs that you have added are gone. I have faced this issue personally, so definitely SnapLogic has to improve in that area. IP whitelisting is currently the top area on my mind for improvement. I don't think of any other areas that need improvement right now.

The learning curve for advanced use cases with SnapLogic could be improved. While basic pipelines are easy, more complex transformations and debugging can still be a bit challenging. Another improvement would be in error messages and debugging clarity. Sometimes the errors are not very descriptive, which makes troubleshooting take longer than expected. Performance tuning and visibility can be better for large-scale pipelines, as having more granular control and deeper insights into execution performance would really help. One improvement would be version control and collaboration. While SnapLogic supports reuse, tighter integration with better change tracking would make team collaboration smoother, especially in larger environments.

One area would be improving the monitor. Another few things I would mention is that I would like better logs to debug in case of any issues, other than just understanding from the pipeline statistics level. I would appreciate some additional logs where on a need basis, I can enable the slots and the traces will be available at the server level so that I can capture the end-to-end what is happening, such as request, response, and things. This feature would be really helpful at the time of critical P1 issues, which I can use. As I mentioned earlier, with few corrections in the monitor and logging and few other areas that I correct and improve, SnapLogic will be much more scalable than others. This is the reason I have given this rating.

I find the learning curve for complex transformations, the debugging experience, performance optimization visibility, cost for small teams, and version control and DevOps integration could be improved in SnapLogic. The improvements needed for SnapLogic include expanded AI agent support, SnapGPT or prompt user enhancements, improved monitoring and observability, expanded connectivity and snaps, Git integrations and DevOps improvements, and platform APIs and automation.

While SnapLogic is powerful, there are several areas for improvement that could enhance user experience. Version control remains an area needing attention as it currently lacks effective features. Debugging complex pipelines can be painful, especially when dealing with deeply nested structures, making it difficult to trace data lineage across pipelines. Improvements in centralized execution and trace visualizations are also necessary. Furthermore, compared to other code-based tools, there is room for advancement in structured transformations, making this a critical area for improvement.

The latency is the biggest issue across iPaaS. That is the important part. I have worked not only with SnapLogic, but also with MuleSoft, Dell Boomi, and Jitterbit. They are all fine, but they have different kinds of functionality. They all have similar kinds of problems in different domains. But as I mentioned, SnapLogic has a little bit of an edge because of its functionality and inbuilt functions for Fintech, as compared to the others. This I can say firmly. They need to assess themselves. In this day and age, as I mentioned earlier, terabytes of data need to be read and then have a quick turnaround for downstream systems, especially for GenAI or any LLMs. Then they can definitely improve themselves because right now, most things, in fact, the GenAI, ChatGPT, Cloud, Anthropic, and so many others, require data quality with perfection and more precision. But for all of that, we require a data pipeline that can be read without latency and without any delay, for any reason. So if they can improvise that over the cloud, that would be really fantastic and a really good achievement for them. Not only for them, but for the customer as well. Then no matter what, people cannot leave SnapLogic. They need to be there with the snaps. I don't know much about that. I haven't referred to the documentation that much. But support is something that is pretty obviously required, rather than just providing videos. Technical support is required. The roadmap also needs to be very clearly mentioned and specified. Be specific in which domain they are going to do what, if they are coming out with that roadmap. Otherwise, overall, if they are going to improve their entire system as I mentioned earlier, for the reusability concept and the data pipeline concept, then they will definitely do some magic in the future. From the HR point of view, or for HR tech, improvement is required. A couple of connectors are not working with all the relevant APIs, and there is always a restriction in terms of fetching the data. So that is why I chose six. From the Fintech point of view, if you are asking on a scale of one to 10, then I would give it an eight out of 10. It is a huge one. There is always a margin for improvement, so that is why I chose eight. If you talk about HR, sales, or any other domain, there is a significant amount of improvement required.

Some advanced features lack detailed documentation, and debugging complex pipelines could be more intuitive. We have faced a few issues with pipeline throttling, but SnapLogic support was responsive and resolved them quickly.

In the future, SnapLogic can make connectors with LLM connectors which are available recently. They can work on that aspect. Sometimes those connectors have low performance or don't generate accurate results, so that can be improved.

I am quite happy with the solution and do not have specific requirements at the moment. I tend to frequently communicate with SnapLogic to ask for additional features, and they have been responsive. While new features are always welcome, I do not have anything else at the top of my mind currently.

SnapLogic is doing well in evolving with the AI landscape. However, if the AI capabilities and integrations were more intuitive and easy to learn for new users, it would be greatly beneficial. Additionally, improving AI capabilities to be easily recognizable would enhance the user experience.

The biggest issue we have faced in our company with SnapLogic is regarding the EDI format. For instance, suppose if 20 EDIs are shortlisted then SnapLogic will convert and provide only those specific 20 EDIs. If a new version of EDI arrives later or if there is a need to use a different existing EDI version, then you need to get back to SnapLogic and complete the setup before initiating the pipeline.

The problem is that SnapLogic doesn't offer a wide variety of connectors. For example, integrating with Salesforce is not that easy, unlike with MuleSoft, which is much easier and also provides a development kit allowing developers to customize or create their own connectors. But SnapLogic lacks this feature. So, it's not easily customizable. For very complex integrations, like those involving 10 input systems and 20 output systems with a variety of data transformations, I wouldn't recommend SnapLogic. You will feel lost somewhere. When dealing with such complex integrations, you often encounter issues managing large datasets. SnapLogic recommends breaking the process into smaller pieces and then deploying them, but sometimes, our requirements are different and require handling multiple systems simultaneously. In future releases, I would like to see better customization. Currently, we have certain connectors, and if we need to build on top of those— tailor them more closely to our business needs—then customization becomes necessary. But here, that process is a bit tricky. I want it to be simpler.

There is room for improvement in customer service and support. I don't think the support has better knowledge about technologies and tool support. There were lots of times when we had an issue, and it took me quite a long time to explain the problem. I feel like some of the support staff don't know their product well.

My company has a licensed version of SnapLogic, so we have to pay for the product when we use it in our organization. Pentaho and Talend are ETL tools that allow my company to migrate the data within a fraction of a second using open-source products, not the paid version. In SnapLogic, my company uses the product's paid version, so we have to pay for license charges. When compared with other tools, SnapLogic needs to be given a trial version or made available as an open-source product for simple migrations, which can be helpful for many users. In Talend and Pentaho, users can migrate data by just installing the software. I just installed the UI. SnapLogic doesn't provide any on-premises software, so users have only cloud-based software to use. Just giving the email address and creating a password is enough to configure SnapLogic. Tools like Talend and Pentaho need their users to indulge in manual installation since both tools fall under the on-premises software category. That is what is actually called on-premises software. It would be great if SnapLogic office separated software to manage small migrations.

I would like to see more performance-related dashboards, ones that display the cost of a pipeline, for instance. Also, it would be helpful to have management dashboards for overseeing pipelines and connections.

They should expand in terms of features for SaaS-based market requirements in different sectors similar to Dell Bumi, AbLogic, Azure, Google, Oracle, etc. There could be more AI capabilities included in the product.

SnapLogic should have some inbuilt protocol mechanism in order to speed up. Each snap is expensive if requested on a one-off basis; however, it may be wiser not to opt for a review unless absolutely necessary. In such cases, there is no need to pay for the review.

I am looking for more scheduling options. When it comes to scheduling, there are different tools in the market.

The solution could improve its API management.

Ultra Pipelines is great in terms of real-time ingestion and real-time jobs but it needs some adjustment. Also, you get the application and enough sources but they don't really ask the correct questions on how to handle the resources and provide the correct information to our team in order to avoid any system problems. We've been seeing some latencies now and then due to the nodes that we currently have available, and due to the lack of knowledge on our SnapLogic Pipelines. In addition, there are snaps we don't need but that affect the latency of the work and I'd like to see some training videos so our developers can have the best practices on how SnapLogic should be used.

What could be improved in SnapLogic is that it was not capable in terms of processing a large number of datasets, but at that point, SnapLogic was evolving. It didn't give a lot of Snaps. I heard recently there are a lot of Snaps getting added and the solution was being enhanced, particularly to connect different data sources. When I was working with SnapLogic six months to one year back, I faced the issue of it not being capable of handling a huge volume of datasets or didn't have much of Snaps, and that was the drawback. If there is any large number of data sets, that's based on or depends on your configuration. If it is a huge volume of data, other traditional ETL tools such as Informatica and Talend can process millions and billions of records, while in SnapLogic, the Snaplex fails or it returns an error in terms of processing that huge volume of data. Informatica, Talend, or any other ETL tool can run for hours in terms of jobs, while SnapLogic jobs fail when the threshold is reached. SnapLogic isn't able to withstand processing, but I don't know if that's still an issue at present, because the solution is getting enhanced and it's been more than six months to one year since I last worked with SnapLogic. There are now a lot of Snaps getting added to the solution, and if it can overcome the limitations I mentioned, SnapLogic could be the go-to tool because currently, it's not being used as much in organizations. It's being used comparatively less compared to other retail tools.

Right now, they have reduced their window period. You will have periodical upgrades from them. Previously, the application used to be down for a while. If you have any real-time pipelines, you need a bridge which is a little costlier solution. You'd need to have a separate server for batching, due to these issues. Otherwise, it'll go down. Your scheduler, you can rerun it, however, in real-time, your business cannot wait. That means you'll maintain a separate server. However, they have reduced this significantly, with downtime down to 15 minutes from two hours. However, if they could further reduce it, that would still be better. We'd like zero downtime in the future. In the past two years, we have seen tremendous growth. We expect very good things in the near future. They, however, do not have their own cloud platform for their iPaaS in the same way AWS or Azure does.

There is room for improvement of API management and the dashboard for task execution. However, they have made some nice improvements in that regard recently. Allowing external services to connect to API management would be great as well.

We'd like there to be more ways for users to get more comfortable and have more experience with the solution to make it easier to use. It would help with referring issues, for example. When we get error messages, I'd like them to be more simplified and easier to understand.

One of the areas for improvement in SnapLogic is that the connectors for some of the applications should be more available in terms of testing in the dev environment. Another area for improvement is that the logging should be standardized, for example, the integration with an ELK stack should be required out-of-the-box, so you can ship the log and have it in the ELK stack. There should be integration with ELK stack for the log shipping. An additional feature I'd like to see in SnapLogic is the functionality of building streaming applications, for example the Pub/Sub model. SnapLogic currently supports Kafka and other streaming applications, but there's more focus on building streaming pipelines now, so if that feature is there, it'll be very helpful.

One area for improvement in SnapLogic is the transparency in the flow of data. It needs to have more transparency. Right now, users only have a preview option at the end of any job flow, so at the end of any Snap Pack, there is a data preview option that lets you review the data and see how it's moving. What would make the solution better is more debugging and more access to change data from the preview panel or more functionality in terms of the preview option. What I'd like to see in the next release of SnapLogic is better debugging of errors. Another feature I'd like to see in the solution is a more in-depth data preview because currently, it just shows you what happened to the data, rather than giving you a preview of how the data gets transformed.

The API monitoring could still be improved upon. The dashboards regarding scheduled tasks need further improvement. It's not user-friendly. If you want to find out which task was executed, you have no good dashboards for monitoring. There are no user-friendly charts where you can check which API was called, and how many times or if there was a peak of requests to a certain API.

The stability can be improved. The problem with Snaplogic is convincing the management. SnapLogic is a technical product, which could be the reason that convincing the management is so difficult. High-volume integration and the number of dependencies should be lower. The EPA requires a lot of customization. Many other solutions are programming-oriented, which require a lot of customization. The support is the most important improvement they could make. In the next release, I would like better support, more online tutorials, online certification courses, and better marketing to help, geared to large businesses and showing them how this solution can meet their requirements. Also, more integration would be helpful.

The solution isn't ideal for complex processing or logic. We use another solution for that. I'm not sure if this solution can integrate with Docker, but if it can't they should work to make this possible as Docker makes it easy to get huge amounts of data and has very good policies.