Tweet
Hello everyone. I wanted to post a couple things. The first is that the Family Finder matching should now be corrected. If you are still missing any matches please let me know or fill out the contact form so it can be looked into on an individual basis.
The other thing is that there was an email that went out to group admins explaining some changes and I wanted to let you know as well. I have included the greater part of it below:
We appreciate your patience with us through the recent bout of technical difficulties we’ve experienced. We know the last couple of weeks have been challenging with the connectivity issues and other bumps in the road, but we assure you we’re working to correct those issues and lay a foundation on which future growth can be built.
Family Tree DNA is unique in the market since, as you know, we’re the only company that offers a full suite of DNA testing for genealogy. Maintaining and improving those discrete pipelines is a challenge, but it’s one we embrace and are committed to meeting.
The most recent and obvious issues have involved matches disappearing and reappearing. We assure you that in those situations, the data from the lab were there and correct. The problem involved the "cache"--a temporary storage location intended to speed up access. It’s similar to your browser’s cache, but at the server level.
The team is first focusing on resolving those caching issues, since they impact the website’s speed and stability, which is the other major issue that has affected everyone.
Here’s a list of the primary corrective actions we’re taking:
We are working to reduce human error by automating our QC processes. We’ve added more QC checks into our processes and will be adding more as needed.
We are upgrading our servers to be able to handle the increased load we are experiencing. In December we doubled our capacity, but more is already needed. These upgrades may take a few weeks to complete.
We are re-organizing our database to distribute the load among all test types so that processing one type doesn’t affect the performance of the other types. For example, uploading a large batch of mtDNA results shouldn’t cause noticeable slowdowns on the site. This network reorganization will take about 2-3 months.
We are moving our results process to a "pipeline" using small, discrete, automated steps. This method will make it much easier to find and fix problems, and will increase processing speed. This will be done in phases over the next 2-3 months. When it is complete, we expect much better consistency and quality.
We are working on new automated pipelines for each DNA line (Y-STR, Big Y, mtDNA, Family Finder) designed to reduce human error. Each pipeline has new quality control checks built in. As we complete each new pipeline, we'll be auditing and re-running existing results to ensure that they are correct, and ensuring that new results go through the added qc checks. This doesn’t mean we haven’t always had QC in place - just that it’s not sufficient for the volume of tests we’re running and will continue to run.
To summarize: in the past, some of our processes relied on manual intervention, which can lead to errors. Increased levels of automation remove that element, allowing us to do a better, more consistent job in the delivery of results. Automation won't make it perfect, but it will help to ensure that we are doing things the same way every time, and that proper QA is done when we do have to fix problems.
-Darren
Family Tree DNA
The other thing is that there was an email that went out to group admins explaining some changes and I wanted to let you know as well. I have included the greater part of it below:
We appreciate your patience with us through the recent bout of technical difficulties we’ve experienced. We know the last couple of weeks have been challenging with the connectivity issues and other bumps in the road, but we assure you we’re working to correct those issues and lay a foundation on which future growth can be built.
Family Tree DNA is unique in the market since, as you know, we’re the only company that offers a full suite of DNA testing for genealogy. Maintaining and improving those discrete pipelines is a challenge, but it’s one we embrace and are committed to meeting.
The most recent and obvious issues have involved matches disappearing and reappearing. We assure you that in those situations, the data from the lab were there and correct. The problem involved the "cache"--a temporary storage location intended to speed up access. It’s similar to your browser’s cache, but at the server level.
The team is first focusing on resolving those caching issues, since they impact the website’s speed and stability, which is the other major issue that has affected everyone.
Here’s a list of the primary corrective actions we’re taking:
We are working to reduce human error by automating our QC processes. We’ve added more QC checks into our processes and will be adding more as needed.
We are upgrading our servers to be able to handle the increased load we are experiencing. In December we doubled our capacity, but more is already needed. These upgrades may take a few weeks to complete.
We are re-organizing our database to distribute the load among all test types so that processing one type doesn’t affect the performance of the other types. For example, uploading a large batch of mtDNA results shouldn’t cause noticeable slowdowns on the site. This network reorganization will take about 2-3 months.
We are moving our results process to a "pipeline" using small, discrete, automated steps. This method will make it much easier to find and fix problems, and will increase processing speed. This will be done in phases over the next 2-3 months. When it is complete, we expect much better consistency and quality.
We are working on new automated pipelines for each DNA line (Y-STR, Big Y, mtDNA, Family Finder) designed to reduce human error. Each pipeline has new quality control checks built in. As we complete each new pipeline, we'll be auditing and re-running existing results to ensure that they are correct, and ensuring that new results go through the added qc checks. This doesn’t mean we haven’t always had QC in place - just that it’s not sufficient for the volume of tests we’re running and will continue to run.
To summarize: in the past, some of our processes relied on manual intervention, which can lead to errors. Increased levels of automation remove that element, allowing us to do a better, more consistent job in the delivery of results. Automation won't make it perfect, but it will help to ensure that we are doing things the same way every time, and that proper QA is done when we do have to fix problems.
-Darren
Family Tree DNA
C'mon guys! It's been almost a month now
Comment