Methods: First, to estimate the burden of COVID in the area, we conducted a descriptive analysis of data on COVID-19 deaths for Collier County from May-August 2020. We then surveyed a cross-sectional, randomized representative sample of 318 adults living in Immokalee from March-November 2020 to assess socio-demographics, workplace conditions, sources of information, ability to follow guidelines, and experiences with testing and contact tracing programs. Results were compared across language groups.
Findings: Average excess mortality in Collier County was 108%. The majority surveyed in Immokalee had socio-demographic factors associated with higher COVID risk. Non-English speakers had higher workplace risk due to less ability to work from home. Haitian Creole speakers were less likely to be tested, though all participants were willing to get symptomatic testing and quarantine. Those participants who tested positive or had COVID-19 exposures had low engagement with the contact tracing program, and Spanish-speakers reported lower quality of contact tracing than English speakers.
Conclusions: The community of Immokalee, FL is a vulnerable population that suffered disproportionate deaths from COVID-19. This study reveals language inequities in COVID testing and contact tracing that should be targeted in future pandemic response in Immokalee and other migrant farmworker communities.
Florida houses one of the largest populations of MSFW in the country, and though the plight of MSFW during the COVID-19 pandemic has gained media attention, there is minimal data around the impact of the pandemic on MSFW communities. Specifically, though it is known that MSFW are a high-risk group, there is no data on the true death rates from COVID-19 for MSFW and there is little information regarding specific workplace conditions, sources of COVID-19 information, ability to follow guidelines, nor their experiences with testing and contact tracing. All of this information is key for improving public health programming.
In this study, we first aimed to estimate the impact of COVID-19 on Immokalee, FL by calculating excess mortality from publicly available data. Then, we conducted a household survey to collect sociodemographic information and assess community experiences with workplace conditions, access to testing, sources of information, and the DOH contact tracing program. Given the linguistic diversity in Immokalee and MSFW populations in general, and limited English proficiency as a known barrier to care, we evaluated for differences in experience by language.
When asked whether participants would utilize testing resources if exposed to COVID-19, a large share of Spanish (89.5%), Haitian Creole (91.4%) and English speakers (87.3%) reported they would be willing to be tested. Most English (53.4%), Spanish (65.4%), and Haitian Creole speakers (66.7%) reported that they would be willing to isolate themselves in a temporary shelter if necessary and they were unable to do so inside their own home.
In this first study of COVID-19 impact on a MSFW population, we found high excess mortality and high COVID-19 risk, with low testing and contact tracing rates and multiple language-based disparities despite many actions from a coalition of community organizations. The 108% excess mortality rate in Collier County calculated in this study is extremely high. In comparison, average excess mortality in Florida was estimated to be 15.5% from March to September, with a peak of 38.1% in August , while nationwide data showed an average of 18.5% excess deaths from March through the end of July . While we cannot discern how many of these excess deaths were specifically from Immokalee, we know it is an especially vulnerable community within the county, as demonstrated by our data. With an average household size of ~4 people with shared bathrooms, over 20% with food and housing insecurity, and a preponderance of essential workers, the risk of COVID-19 infection was and continues to be high in this community . For non-English speakers, that risk is even higher, as our data show they are less frequently able to work from home.
Confronted by this excess disease burden in a highly vulnerable population, the stakeholder groups came together rapidly and worked towards a COVID-19 response. Through their combined efforts, a new system for testing, quarantine and contact tracing was put into place. Our data show the successes and challenges of setting up such a system in Immokalee, with lessons relevant to MSFW populations across the U.S.
The initial testing system was based on appointments scheduled via an English-only online portal. While most surveyed residents indicated willingness to test and quarantine, the data show marked language disparities in testing, with significantly lower testing rates in Haitian Creole speakers. The HCN helped address this disparity a few months into testing by initiating rapid antigen testing at mobile field sites run by Haitian Creole- and Spanish-speaking staff. Our data show that this change may have facilitated a higher testing rate in the Haitian Creole population; amongst the number of Haitian Creole speakers tested, the majority received tests at these HCN rapid-test sites. However, the data show overall low testing rates for all language groups despite these collective efforts, demonstrating a need for even more accessible testing for predominantly MSFW populations.
These disparities in testing and contact tracing highlight the challenges that the population and coalition of community organizations faced while responding to the high COVID-19 rates in Immokalee. The disparities demonstrate the need for all collaborators in the larger health system to connect more effectively with vulnerable communities like MSFW and other essential workers, especially Haitian Creole speakers who are known to be particularly marginalized.
Overall, this study quantifies the impact of COVID-19 on Immokalee and elucidates the individual, household, and occupational factors that place this community at especially high risk. The data show that despite coordinated efforts from a committed group of collaborators, significant language-based inequities impacted the risk of contracting COVID-19, testing rates, and receiving high-quality contact tracing. These inequities are a proxy for the disproportionate barriers faced by non-English speaking populations in Immokalee to access care.
Authors are thankful to Dr. K. M. Paknikar Director, ARI and Agharkar Research Institute, Pune, India and IIT Guwahati, for providing infrastructure. Ninad V. Puranik, Gaurav Bhatt, and Dixcy Jaba Sheeba John Mary are thankful for their fellowship. We would like to acknowledge the extreme support of Mr. Vinod Devaraji, to understand the Maestro software from Schrodinger.
Both works above, however, had one crucial aspect in common. We focused on onetype of datapath, exclusively, namely the kernel datapath installed by theunderlying system's own packet manager. In many real-world (production)environments, administrators simply rely on the built-in software tools toinstall applications to reduce or even completely avoid all the crux aroundmanual installations and compilations from source code, e.g., via apt-getinstall openvswitch-common in Debian-based Linux distributions. Even though inmost of the cases, we eventually end up having the same application withnegligible (performance) difference, when applications also have modulessupplied by the underlying kernel (e.g., in the case of Open vSwitch since theLinux 3.3 kernel debut in 2012 ), there can be significant deviations amongthe implementations. In particular, as it turned out after the discussions(with some of the OVS developers) during our previous talks, (i) the kernelnetworking stack developers do not prefer exact flow caching; therefore, thekernel datapath of OVS lacks the first-level Exact Match Cache (EMC). Thismeans that the whole fast-path only comprises the MFC, thereby making TSE moreefficient. On the other hand, (ii) while the userspace datapath provided byIntel's DPDK significantly improves the packet processing performance (byavoiding context-switching, interrupt-based packet handling, and theside-effects of OS schedulers), it essentially shares the same code base, andmost parts of the algorithms are implemented according to the same originaldesign.
To counter this ranking algorithm, we propose TSE 2.0, which by letting sometuples expire and re-spawning them by carefully switching the original TSEattack on and off, keeps the ranking algorithm busy. Thus, eventually, TSE 2.0causes a complete denial-of-service (DoS) for the users of the same softwareswitch. Furthermore, we propose TSE 2.1 against OVS-DPDK running on multiplecores, wherein we slightly increase the attack rate of TSE 2.0, but, at thesame time, we carefully adjust the packet sending sequence to achieve the sameresults as with TSE 2.0. We experimentally show that TSE 2.1 can still mount alow-rate DoS attack as long as OVS-DPDK is running on less than five cores.
OVSDB was originally designed to store configuration for Open vSwitch daemon.It was intended to handle fairly small amount of data that covers networkconfiguration for a single host. But, at the same time, it was designed tobe flexible and easy to talk with via OVSDB management protocol (RFC7047).Today OVN becomes a primary user of OVSDB and this rises new challenges inthe area of performance and scale. This talk is about debugging issues andperformance testing of OVSDB (and other components) in your OVN deployments.Just record the problematic use-case or a production workload and debug itwhile sitting at home and using only power of your own laptop.
This presentation is about integrating two very powerful but dissimilarnetworking technologies. We want to simultaneously harness the power of theubiquitous presence of the OVS-like software switches, and the flexibility ofprotocol and platform-independence that the P4 language provides. We mustintegrate the two in a meaningful way to combine their strengths and overcometheir limitations. Our approach for this integration involves introducingp4proto as a parallel to ofproto within OVS, that maps Openflow tables to P4tables, and establishes a compiler-based framework to efficiently map the P4tables to any device's physical tables. We will also talk about converting OVSconfigurations into P4 table configurations for LAG, Mirroring, CT and S-Flow,to easily program these features on P4 based data planes directly, as comparedto the current flow driven implementations of OVS. 2b1af7f3a8